As a Member of Technical Staff (Research) at Trajectory, you will design and build the post‑training stack that lets our customers’ models continually learn from real production workflows. You will own end‑to‑end experiments across data, training, and evaluation: shaping telemetry into learnable signals, training and serving custom LLMs and agents, and designing novel algorithms. You’ll work directly with the founders and design partners, make foundational technical and product decisions, and help define both the culture and architecture of a company whose core mission is to enable continually learning intelligence.
Experience training or serving large language models
Experience building RL environments or evaluation suites for LLMs
Proficiency in PyTorch, JAX, or similar ML frameworks
Strong experimental design skills
Master’s / research experience
Background in pre- or post-training RL for LLMs
Experience with high-performance computing or large-scale clusters
Contributions to open-source ML research or infrastructure
Demonstrated technical creativity (research, OSS, side projects)
PhD (last year of PhD), working on AI / RL open source projects
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox.
","descriptionPlain":"JOB DESCRIPTION\n\nAs a Member of Technical Staff (Research) at Trajectory, you will design and build the post‑training stack that lets our customers’ models continually learn from real production workflows. You will own end‑to‑end experiments across data, training, and evaluation: shaping telemetry into learnable signals, training and serving custom LLMs and agents, and designing novel algorithms. You’ll work directly with the founders and design partners, make foundational technical and product decisions, and help define both the culture and architecture of a company whose core mission is to enable continually learning intelligence.\n\n\n\n\nREQUIRED QUALIFICATIONS\n\n - Experience training or serving large language models\n\n - Experience building RL environments or evaluation suites for LLMs\n\n - Proficiency in PyTorch, JAX, or similar ML frameworks\n\n - Strong experimental design skills\n\n - Master’s / research experience\n\n\n\n\nPREFERRED\n\n - Background in pre- or post-training RL for LLMs\n\n - Experience with high-performance computing or large-scale clusters\n\n - Contributions to open-source ML research or infrastructure\n\n - Demonstrated technical creativity (research, OSS, side projects)\n\n - PhD (last year of PhD), working on AI / RL open source projects\n\n\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox."},{"id":"6f0eb3f0-d042-44ed-be97-eac1608cfab5","title":"Member of Technical Staff - Design Eng","department":"Member of Technical Staff","team":"Member of Technical Staff","employmentType":"FullTime","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-05-26T03:57:57.652+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/6f0eb3f0-d042-44ed-be97-eac1608cfab5","applyUrl":"https://jobs.ashbyhq.com/trajectory/6f0eb3f0-d042-44ed-be97-eac1608cfab5/application","descriptionHtml":"As a Design Engineer at Trajectory, you will craft the product experience that makes continual learning legible and intuitive.
You will own the visualization, interaction, and developer experience surfaces that help customers understand, debug, and trust continually improving models. This includes trajectory visualization, evaluation workflows, monitoring dashboards, and the core product interfaces that sit on top of our leading RL infrastructure.
You’ll work directly with the founders and industry-leading design partners (Series B+ AI vertical product companies) to translate frontier RL systems into a clean, developer-first product experience. You will help define the visual language, interaction patterns, and product philosophy of Trajectory.
We are building the interface for continual learning. This role sets that bar.
Note: Some travel may be required - we work closely with design partners across the world.
Strong frontend or design engineering experience (TypeScript/React or similar)
Experience building polished, production user interfaces
Strong product taste and clear opinions about what makes software feel great - with the craft to back it up
Experience building AI-powered products or agent systems - you actively experiment with new frameworks for fun!
Experience building data visualization or complex technical UIs
Experience with observability or debugging interfaces
Background at high-craft product companies or developer tooling startups
Experience working closely with designers or operating as a design-minded engineer
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox.
","descriptionPlain":"JOB DESCRIPTION\n\nAs a Design Engineer at Trajectory, you will craft the product experience that makes continual learning legible and intuitive.\n\nYou will own the visualization, interaction, and developer experience surfaces that help customers understand, debug, and trust continually improving models. This includes trajectory visualization, evaluation workflows, monitoring dashboards, and the core product interfaces that sit on top of our leading RL infrastructure.\n\nYou’ll work directly with the founders and industry-leading design partners (Series B+ AI vertical product companies) to translate frontier RL systems into a clean, developer-first product experience. You will help define the visual language, interaction patterns, and product philosophy of Trajectory.\n\nWe are building the interface for continual learning. This role sets that bar.\n\nNote: Some travel may be required - we work closely with design partners across the world.\n\n\n\n\n\nREQUIRED QUALIFICATIONS\n\n - Strong frontend or design engineering experience (TypeScript/React or similar)\n\n - Experience building polished, production user interfaces\n\n - Strong product taste and clear opinions about what makes software feel great - with the craft to back it up\n\n - Experience building AI-powered products or agent systems - you actively experiment with new frameworks for fun!\n\n\n\n\nPREFERRED\n\n - Experience building data visualization or complex technical UIs\n\n - Experience with observability or debugging interfaces\n\n - Background at high-craft product companies or developer tooling startups\n\n - Experience working closely with designers or operating as a design-minded engineer\n\n\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox."},{"id":"490e11d6-7a02-4c58-8937-ecfc8811ea19","title":"Member of Technical Staff - Generalist","department":"Member of Technical Staff","team":"Member of Technical Staff","employmentType":"FullTime","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-05-07T00:30:28.751+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/490e11d6-7a02-4c58-8937-ecfc8811ea19","applyUrl":"https://jobs.ashbyhq.com/trajectory/490e11d6-7a02-4c58-8937-ecfc8811ea19/application","descriptionHtml":"As a Generalist Engineer at Trajectory, you will craft the product experience that makes continual learning legible and intuitive.
You will own the SDK, API surfaces, and integration infrastructure that help customers understand, debug, and trust continually improving models. This includes trajectory visualization, evaluation workflows, monitoring dashboards, and the core product interfaces that sit on top of our leading RL infrastructure.
You’ll work directly with the founders and industry-leading design partners (Series B+ AI vertical product companies) to translate frontier RL systems into a clean, developer-first product experience. You will help define the visual language, interaction patterns, and product philosophy of Trajectory.
We are building the interface for continual learning. This role sets that bar.
Strong full-stack or backend engineering experience
Experience building AI-powered products or agent systems - you actively experiment with new frameworks for fun
Experience building developer-facing products or SDKs
Experience with data pipelines, telemetry systems, or event infrastructure
Ability to move quickly in a zero-to-one environment
Experience building observability, tracing, or monitoring systems
Experience at early-stage developer tooling startups
Strong product instincts and ability to work directly with customers
Experience owning SDK design and developer ergonomics
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox.
","descriptionPlain":"JOB DESCRIPTION\n\nAs a Generalist Engineer at Trajectory, you will craft the product experience that makes continual learning legible and intuitive.\n\nYou will own the SDK, API surfaces, and integration infrastructure that help customers understand, debug, and trust continually improving models. This includes trajectory visualization, evaluation workflows, monitoring dashboards, and the core product interfaces that sit on top of our leading RL infrastructure.\n\nYou’ll work directly with the founders and industry-leading design partners (Series B+ AI vertical product companies) to translate frontier RL systems into a clean, developer-first product experience. You will help define the visual language, interaction patterns, and product philosophy of Trajectory.\n\nWe are building the interface for continual learning. This role sets that bar.\n\n\n\n\n\nREQUIRED QUALIFICATIONS\n\n - Strong full-stack or backend engineering experience\n\n - Experience building AI-powered products or agent systems - you actively experiment with new frameworks for fun\n\n - Experience building developer-facing products or SDKs\n\n - Experience with data pipelines, telemetry systems, or event infrastructure\n\n - Ability to move quickly in a zero-to-one environment\n\n\nPREFERRED\n\n - Experience building observability, tracing, or monitoring systems\n\n - Experience at early-stage developer tooling startups\n\n - Strong product instincts and ability to work directly with customers\n\n - Experience owning SDK design and developer ergonomics\n\n\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox."},{"id":"a9059736-e688-4d6e-91ba-5b968e104bad","title":"Member of Business Staff","department":"Member of Business Staff","team":"Member of Business Staff","employmentType":"FullTime","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-09-10T23:41:09.464+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/a9059736-e688-4d6e-91ba-5b968e104bad","applyUrl":"https://jobs.ashbyhq.com/trajectory/a9059736-e688-4d6e-91ba-5b968e104bad/application","descriptionHtml":"Trajectory is hiring a Member of Business Staff to help build the business around our continual-learning platform. You will work directly with the founders across go-to-market, strategic partnerships, customer success, pricing, external communications, and business operations.
Our market, product, and commercial model are still being defined. We need someone who can understand technical customers deeply, expand the space of possible opportunities, turn ambiguous problems into systems, communicate complex ideas with exceptional clarity, and personally execute the work required to move them forward.
Make each design partner a priority through excellent preparation, communication, and follow-through.
Run timelines, meeting cadences, materials, onboarding, and cross-functional coordination.
Track customer usage and feedback, identify blockers, and bring actionable insights into product and research planning.
Build lightweight support, renewal, reference, and expansion processes.
Distinguish customer-specific requests from reusable capabilities that should become part of the product.
Run price discovery with customers and map how budgets and buying decisions work.
Help design pricing and packaging across platform access, training and evaluation usage, inference, and delivered outcomes.
Structure and negotiate pilots, contracts, credits, commitments, renewals, and expansions.
Build models that connect customer value, compute costs, product usage, and long-term economics.
Develop partnerships with model providers, cloud and compute vendors, data partners, and distribution channels.
Lead opportunities with organizations such as cloud providers, inference platforms, and ecosystem partners.
Negotiate commercial terms, credits, capacity, co-marketing, and customer-referral structures.
Protect Trajectory’s direct customer relationships, product position, and long-term strategic leverage.
Sharpen how Trajectory explains its product, market, and differentiation.
Help define the language for the continual-learning category.
Turn technical research, product progress, and customer outcomes into credible blog posts, case studies, launch announcements, website copy, and co-marketing opportunities.
Talk to researchers, customers, and partners; find the important story inside complex work; and carry it from rough notes to polished publication.
Build and manage an editorial calendar in partnership with the founders, research, product, and design teams.
Partner with the team on events, customer moments, technical storytelling, social content, and sales enablement.
Develop thoughtful distribution strategies that reach technical buyers, partners, candidates, and industry voices—not broad marketing for its own sake.
Define (and refine) our ideal customer profiles.
Build target-account maps, identify internal champions, and create intentional outbound strategies.
Own pipeline management, qualification, follow-ups, and deal momentum.
Develop the sales process from first conversation through technical evaluation, onboarding, and expansion.
Create systems and metrics for understanding conversion, adoption, deployment depth, retention, and expansion.
Develop a deep understanding of each customer’s agent, data, evaluation process, technical constraints, and business goals.
Diagnose where model or harness improvements create meaningful customer value.
Translate technical capabilities and research results into clear proposals, decks, demos, one-pagers, and decision documents.
Define success criteria for pilots and ensure that each engagement has a credible path to production.
Help customers make clear decisions rather than allowing promising conversations to drift.
You are a systems thinker who turns ambiguous goals into clear decisions, processes, and execution.
You are intellectually curious and can develop a real understanding of technical products, model training, evaluations, agents, and infrastructure.
You are an exceptional writer and editor. You can make complex technical work precise, compelling, and memorable without flattening its substance.
You communicate clearly in person, especially with technical founders, researchers, product leaders, and executives.
You have strong commercial judgment without sounding or behaving like a conventional salesperson.
You are highly proactive and low ego.
You are excellent at customer discovery: you can separate stated requests from underlying problems and business value.
You can operate across strategy and details; one hour defining a market and the next writing a deck or blog post, negotiating a contract, or fixing a broken process.
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. Raised $60M from Sequoia, Conviction, along with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, Dropbox.
","descriptionPlain":"THE ROLE\n\nTrajectory is hiring a Member of Business Staff to help build the business around our continual-learning platform. You will work directly with the founders across go-to-market, strategic partnerships, customer success, pricing, external communications, and business operations.\n\nOur market, product, and commercial model are still being defined. We need someone who can understand technical customers deeply, expand the space of possible opportunities, turn ambiguous problems into systems, communicate complex ideas with exceptional clarity, and personally execute the work required to move them forward.\n\n\nWHAT YOU’LL DO\n\n\nOWN CUSTOMER SUCCESS AND EXPANSION\n\n - Make each design partner a priority through excellent preparation, communication, and follow-through.\n\n - Run timelines, meeting cadences, materials, onboarding, and cross-functional coordination.\n\n - Track customer usage and feedback, identify blockers, and bring actionable insights into product and research planning.\n\n - Build lightweight support, renewal, reference, and expansion processes.\n\n - Distinguish customer-specific requests from reusable capabilities that should become part of the product.\n\n\nDEVELOP PRICING AND THE BUSINESS MODEL\n\n - Run price discovery with customers and map how budgets and buying decisions work.\n\n - Help design pricing and packaging across platform access, training and evaluation usage, inference, and delivered outcomes.\n\n - Structure and negotiate pilots, contracts, credits, commitments, renewals, and expansions.\n\n - Build models that connect customer value, compute costs, product usage, and long-term economics.\n\n\nCREATE STRATEGIC PARTNERSHIPS\n\n - Develop partnerships with model providers, cloud and compute vendors, data partners, and distribution channels.\n\n - Lead opportunities with organizations such as cloud providers, inference platforms, and ecosystem partners.\n\n - Negotiate commercial terms, credits, capacity, co-marketing, and customer-referral structures.\n\n - Protect Trajectory’s direct customer relationships, product position, and long-term strategic leverage.\n\n\nPOSITIONING, WRITING, AND EXTERNAL COMMUNICATIONS\n\n - Sharpen how Trajectory explains its product, market, and differentiation.\n\n - Help define the language for the continual-learning category.\n\n - Turn technical research, product progress, and customer outcomes into credible blog posts, case studies, launch announcements, website copy, and co-marketing opportunities.\n\n - Talk to researchers, customers, and partners; find the important story inside complex work; and carry it from rough notes to polished publication.\n\n - Build and manage an editorial calendar in partnership with the founders, research, product, and design teams.\n\n - Partner with the team on events, customer moments, technical storytelling, social content, and sales enablement.\n\n - Develop thoughtful distribution strategies that reach technical buyers, partners, candidates, and industry voices—not broad marketing for its own sake.\n\n\nBUILD THE GO-TO-MARKET MOTION\n\n - Define (and refine) our ideal customer profiles.\n\n - Build target-account maps, identify internal champions, and create intentional outbound strategies.\n\n - Own pipeline management, qualification, follow-ups, and deal momentum.\n\n - Develop the sales process from first conversation through technical evaluation, onboarding, and expansion.\n\n - Create systems and metrics for understanding conversion, adoption, deployment depth, retention, and expansion.\n\n\nLEAD DISCOVERY AND PRE-SALES\n\n - Develop a deep understanding of each customer’s agent, data, evaluation process, technical constraints, and business goals.\n\n - Diagnose where model or harness improvements create meaningful customer value.\n\n - Translate technical capabilities and research results into clear proposals, decks, demos, one-pagers, and decision documents.\n\n - Define success criteria for pilots and ensure that each engagement has a credible path to production.\n\n - Help customers make clear decisions rather than allowing promising conversations to drift.\n\n\nWHO YOU ARE\n\n - You are a systems thinker who turns ambiguous goals into clear decisions, processes, and execution.\n\n - You are intellectually curious and can develop a real understanding of technical products, model training, evaluations, agents, and infrastructure.\n\n - You are an exceptional writer and editor. You can make complex technical work precise, compelling, and memorable without flattening its substance.\n\n - You communicate clearly in person, especially with technical founders, researchers, product leaders, and executives.\n\n - You have strong commercial judgment without sounding or behaving like a conventional salesperson.\n\n - You are highly proactive and low ego.\n\n - You are excellent at customer discovery: you can separate stated requests from underlying problems and business value.\n\n - You can operate across strategy and details; one hour defining a market and the next writing a deck or blog post, negotiating a contract, or fixing a broken process.\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. Raised $60M from Sequoia, Conviction, along with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, Dropbox."},{"id":"00cd9b1a-5d48-4ecb-8c53-cd27727448c5","title":"Member of Technical Staff - Intern","department":"Member of Technical Staff","team":"Member of Technical Staff","employmentType":"Intern","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-11-05T19:37:29.926+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/00cd9b1a-5d48-4ecb-8c53-cd27727448c5","applyUrl":"https://jobs.ashbyhq.com/trajectory/00cd9b1a-5d48-4ecb-8c53-cd27727448c5/application","descriptionHtml":"Please choose a role above to view details and apply through its dedicated application form.
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier.
Our research team comes from Deepmind, OpenAI, Meta Superintelligence, and product team from Figma, Apple, Stripe, and Windsurf. We're working with customers like Harvey, Rogo, Mercor, Decagon and Clay, and we've raised 60M from Sequoia, Conviction, Jeff Dean, and Fei Fei Li.
","descriptionPlain":"EXPLORE OUR INTERNSHIP ROLES\n\n\nGENERALIST INTERN https://jobs.ashbyhq.com/trajectory/ceee0ea0-195d-4c58-8b57-4013c81e373c\n\n\nRESEARCH INTERN https://jobs.ashbyhq.com/trajectory/a3fd075c-2480-4567-9bec-e587f2208ac8\n\nPlease choose a role above to view details and apply through its dedicated application form.\n\n\n\n\nABOUT US\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. \n\nOur research team comes from Deepmind, OpenAI, Meta Superintelligence, and product team from Figma, Apple, Stripe, and Windsurf. We're working with customers like Harvey, Rogo, Mercor, Decagon and Clay, and we've raised 60M from Sequoia, Conviction, Jeff Dean, and Fei Fei Li."},{"id":"60480265-9c60-4281-9d09-2ed18268c58a","title":"Member of Technical Staff - ML Infra","department":"Member of Technical Staff","team":"Member of Technical Staff","employmentType":"FullTime","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-10-06T22:53:37.591+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/60480265-9c60-4281-9d09-2ed18268c58a","applyUrl":"https://jobs.ashbyhq.com/trajectory/60480265-9c60-4281-9d09-2ed18268c58a/application","descriptionHtml":"As an ML Infrastructure Engineer at Trajectory, you will develop algorithms and own the innermost loop of our continual learning platform: the ML training stack!
Our team’s goal is to improve the training stack by 10× every month across GPU scale, model size, throughput, memory efficiency, caching, and latency.
This role spans training, inference, and kernels. Bring deep expertise in at least one area and curiosity across the stack. We’ll shape your initial ownership around your strengths and have high opportunities for you to work across the whole ML training platform.
Training
Build and optimize distributed training and RL infrastructure, including rollout execution, GPU scheduling, training parallelisms, checkpointing, and recovery. Improve training experimentation throughput and shorten research iteration cycles is a key part of training 100B+ parameter models here.
Inference
Optimize serving for production agents and training rollouts. Improve batching, scheduling, and KV-cache management while balancing latency, throughput, cost, and model quality.
Kernels and runtimes
Develop and optimize GPU kernels and runtimes using CUDA, Triton, or comparable tools. Improve memory use and execution efficiency, preserving numerical correctness and verifying gains in real workloads.
Across all three
Build reproducible benchmarks, observability, and automated research workflows that propose changes, run experiments, and validate improvements. Work with researchers to turn new algorithms into reliable systems across training, inference, and kernels.
We’re building one of the world’s best continual learning loops for ML infrastructure: agents that can propose optimizations, run experiments, measure gains, and learn from experiments across training, inference, and kernels. Come join us to accelerate continual learning’s inner loop!
Strong fundamentals in distributed systems, networking, storage, and failure recovery, with experience shipping demanding systems.
Required deep specialization in one of training infrastructure, inference systems, or GPU kernels, supported by systems built or measurable optimizations delivered.
Strong Python skills and languages relevant to your specialty, such as C++, CUDA, or Triton.
Understanding of PyTorch, JAX, or comparable framework internals, with strong profiling and debugging skills.
Ownership and clear communication: work closely with researchers and deliver measurable performance gains while preserving correctness and reliability.
We value demonstrated capability over credentials.
Hands-on experience with training and RL stacks such as Miles, SkyRL, Prime Intellect’s verifiers, or comparable systems. Depending on your specialty, experience with vLLM, SGLang, collective communication, or ML compilers is also valuable.
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier.
Our research team comes from Deepmind, OpenAI, Meta Superintelligence, and product team from Figma, Apple, Stripe, and Windsurf. We're working with customers like Harvey, Rogo, Mercor, Decagon and Clay, and we've raised $60M from Sequoia, Conviction, Jeff Dean, and Fei Fei Li.
","descriptionPlain":"JOB DESCRIPTION\n\nAs an ML Infrastructure Engineer at Trajectory, you will develop algorithms and own the innermost loop of our continual learning platform: the ML training stack!\n\nOur team’s goal is to improve the training stack by 10× every month across GPU scale, model size, throughput, memory efficiency, caching, and latency.\n\nThis role spans training, inference, and kernels. Bring deep expertise in at least one area and curiosity across the stack. We’ll shape your initial ownership around your strengths and have high opportunities for you to work across the whole ML training platform.\n\n\nWHAT YOU WILL WORK ON\n\nTraining\n\nBuild and optimize distributed training and RL infrastructure, including rollout execution, GPU scheduling, training parallelisms, checkpointing, and recovery. Improve training experimentation throughput and shorten research iteration cycles is a key part of training 100B+ parameter models here.\n\nInference\n\nOptimize serving for production agents and training rollouts. Improve batching, scheduling, and KV-cache management while balancing latency, throughput, cost, and model quality.\n\nKernels and runtimes\n\nDevelop and optimize GPU kernels and runtimes using CUDA, Triton, or comparable tools. Improve memory use and execution efficiency, preserving numerical correctness and verifying gains in real workloads.\n\nAcross all three\n\nBuild reproducible benchmarks, observability, and automated research workflows that propose changes, run experiments, and validate improvements. Work with researchers to turn new algorithms into reliable systems across training, inference, and kernels.\n\n\nCONTINUAL LEARNING FOR ML INFRASTRUCTURE\n\nWe’re building one of the world’s best continual learning loops for ML infrastructure: agents that can propose optimizations, run experiments, measure gains, and learn from experiments across training, inference, and kernels. Come join us to accelerate continual learning’s inner loop!\n\n\nREQUIRED QUALIFICATIONS\n\n - Strong fundamentals in distributed systems, networking, storage, and failure recovery, with experience shipping demanding systems.\n\n - Required deep specialization in one of training infrastructure, inference systems, or GPU kernels, supported by systems built or measurable optimizations delivered.\n\n - Strong Python skills and languages relevant to your specialty, such as C++, CUDA, or Triton.\n\n - Understanding of PyTorch, JAX, or comparable framework internals, with strong profiling and debugging skills.\n\n - Ownership and clear communication: work closely with researchers and deliver measurable performance gains while preserving correctness and reliability.\n\nWe value demonstrated capability over credentials.\n\n\nPREFERRED\n\nHands-on experience with training and RL stacks such as Miles, SkyRL, Prime Intellect’s verifiers, or comparable systems. Depending on your specialty, experience with vLLM, SGLang, collective communication, or ML compilers is also valuable.\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier.\n\nOur research team comes from Deepmind, OpenAI, Meta Superintelligence, and product team from Figma, Apple, Stripe, and Windsurf. We're working with customers like Harvey, Rogo, Mercor, Decagon and Clay, and we've raised $60M from Sequoia, Conviction, Jeff Dean, and Fei Fei Li."},{"id":"d54a128d-4b62-40c0-8319-6b6a0f2f39c8","title":"Founding Recruiter","department":"Member of Ops Staff","team":"Member of Ops Staff","employmentType":"FullTime","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-09-29T23:42:19.447+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/d54a128d-4b62-40c0-8319-6b6a0f2f39c8","applyUrl":"https://jobs.ashbyhq.com/trajectory/d54a128d-4b62-40c0-8319-6b6a0f2f39c8/application","descriptionHtml":"Trajectory’s mission to build the platform for continual learning is ambitious, and requires exceptional talent to make it happen. We're hiring a Founding Recruiter to design the operating system for how we find outliers, assess their fit for Trajectory, and convince them to join. This is a full-stack recruiting role: you own the entire funnel end to end, from sourcing to interviewing to closing.
The best talent requires concerted time and effort throughout the recruiting process. We need someone who takes a first-principles, engineering mindset to talent to make the unscalable scalable, and creates new ways to hire exceptional people at speed, and is personally the person they meet.
Design how we identify exceptional people, assess their fit, and attract them to Trajectory.
Design an efficient interview pipeline that can go from first intro to offer in the shortest possible time
Build the systems that keep someone in Trajectory’s orbit (socials, coffee chats, intros) and encode scalable context so we can tailor our process to best fit what candidates our looking for
Run hand-sourcing as an always-on motion: talent mapping against target companies and labs, referral asks from every new hire and investor.
Go where the best people are and aren't looking for a job — open source, papers, conferences, competition circuits, campus hubs, X and GitHub follow graphs.
Build a personalized closing plan for every candidate we want, grounded in their motivations, blockers, timeline, and mutuals.
Work directly with the founders to go the unscalable last mile for every candidate, from the beginning of the process all the way to closing
You are a systems thinker who turns ambiguous processes into clear decisions, processes, execution, and automation.
You go the last mile. Finding and closing talent means hosting tailored events, identifying thoughtful gifts for candidates, and breaking process to move the needle.
You have taste for talent in technical and unusual roles. You can tell a polished, recruiter-shaped candidate from someone who will actually do the work.
You are an excellent communicator. You build trust quickly, convey context clearly, and adapt your message to candidates, founders, and teammates without losing precision.
You are metrics-driven. You define the measures that matter across the recruiting funnel, inspect conversion, speed, and quality, and use the data to decide what to improve next.
You are AI-pilled. Claude Code to build out dashboards and automations. Grok Bot to streamline comms and outreach. Any process you do twice can be systemized with AI.
You are highly proactive and low ego.
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Trajectory unlocks the signal already sitting in product use so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from DeepMind, OpenAI, Meta, Apple, Amazon, and Scale. Raised $60M from Sequoia, Conviction, along with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox.
","descriptionPlain":"THE ROLE\n\nTrajectory’s mission to build the platform for continual learning is ambitious, and requires exceptional talent to make it happen. We're hiring a Founding Recruiter to design the operating system for how we find outliers, assess their fit for Trajectory, and convince them to join. This is a full-stack recruiting role: you own the entire funnel end to end, from sourcing to interviewing to closing.\n\nThe best talent requires concerted time and effort throughout the recruiting process. We need someone who takes a first-principles, engineering mindset to talent to make the unscalable scalable, and creates new ways to hire exceptional people at speed, and is personally the person they meet.\n\n \n\n\nWHAT YOU'LL DO\n\n\n1. BUILD THE TALENT OPERATING SYSTEM\n\n - Design how we identify exceptional people, assess their fit, and attract them to Trajectory.\n\n - Design an efficient interview pipeline that can go from first intro to offer in the shortest possible time\n\n - Build the systems that keep someone in Trajectory’s orbit (socials, coffee chats, intros) and encode scalable context so we can tailor our process to best fit what candidates our looking for\n\n\n2. SOURCE THE TOP 0.1%\n\n - Run hand-sourcing as an always-on motion: talent mapping against target companies and labs, referral asks from every new hire and investor.\n\n - Go where the best people are and aren't looking for a job — open source, papers, conferences, competition circuits, campus hubs, X and GitHub follow graphs.\n\n\n3. CLOSE\n\n - Build a personalized closing plan for every candidate we want, grounded in their motivations, blockers, timeline, and mutuals.\n\n - Work directly with the founders to go the unscalable last mile for every candidate, from the beginning of the process all the way to closing\n\n\nWHO YOU ARE\n\n - You are a systems thinker who turns ambiguous processes into clear decisions, processes, execution, and automation.\n\n - You go the last mile. Finding and closing talent means hosting tailored events, identifying thoughtful gifts for candidates, and breaking process to move the needle.\n\n - You have taste for talent in technical and unusual roles. You can tell a polished, recruiter-shaped candidate from someone who will actually do the work.\n\n - You are an excellent communicator. You build trust quickly, convey context clearly, and adapt your message to candidates, founders, and teammates without losing precision.\n\n - You are metrics-driven. You define the measures that matter across the recruiting funnel, inspect conversion, speed, and quality, and use the data to decide what to improve next.\n\n - You are AI-pilled. Claude Code to build out dashboards and automations. Grok Bot to streamline comms and outreach. Any process you do twice can be systemized with AI.\n\n - You are highly proactive and low ego.\n\n\n\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Trajectory unlocks the signal already sitting in product use so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from DeepMind, OpenAI, Meta, Apple, Amazon, and Scale. Raised $60M from Sequoia, Conviction, along with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox."},{"id":"a3fd075c-2480-4567-9bec-e587f2208ac8","title":"Member of Technical Staff - Research Intern","department":"Member of Technical Staff","team":"Member of Technical Staff","employmentType":"Intern","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-11-05T19:14:29.217+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/a3fd075c-2480-4567-9bec-e587f2208ac8","applyUrl":"https://jobs.ashbyhq.com/trajectory/a3fd075c-2480-4567-9bec-e587f2208ac8/application","descriptionHtml":"As a Member of Technical Staff (Research) at Trajectory, you will design and build the post‑training stack that lets our customers’ models continually learn from real production workflows. You will own end‑to‑end experiments across data, training, and evaluation: shaping telemetry into learnable signals, training and serving custom LLMs and agents, and designing novel algorithms. You’ll work directly with the founders and design partners, make foundational technical and product decisions, and help define both the culture and architecture of a company whose core mission is to enable continually learning intelligence.
Experience training or serving large language models
Experience building RL environments or evaluation suites for LLMs
Proficiency in PyTorch, JAX, or similar ML frameworks
Strong experimental design skills
Master’s / research experience
Background in pre- or post-training RL for LLMs
Experience with high-performance computing or large-scale clusters
Contributions to open-source ML research or infrastructure
Demonstrated technical creativity (research, OSS, side projects)
PhD (last year of PhD), working on AI / RL open source projects
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox.
","descriptionPlain":"JOB DESCRIPTION\n\nAs a Member of Technical Staff (Research) at Trajectory, you will design and build the post‑training stack that lets our customers’ models continually learn from real production workflows. You will own end‑to‑end experiments across data, training, and evaluation: shaping telemetry into learnable signals, training and serving custom LLMs and agents, and designing novel algorithms. You’ll work directly with the founders and design partners, make foundational technical and product decisions, and help define both the culture and architecture of a company whose core mission is to enable continually learning intelligence.\n\n\n\n\nREQUIRED QUALIFICATIONS\n\n - Experience training or serving large language models\n\n - Experience building RL environments or evaluation suites for LLMs\n\n - Proficiency in PyTorch, JAX, or similar ML frameworks\n\n - Strong experimental design skills\n\n - Master’s / research experience\n\n\n\n\nPREFERRED\n\n - Background in pre- or post-training RL for LLMs\n\n - Experience with high-performance computing or large-scale clusters\n\n - Contributions to open-source ML research or infrastructure\n\n - Demonstrated technical creativity (research, OSS, side projects)\n\n - PhD (last year of PhD), working on AI / RL open source projects\n\n\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox."},{"id":"ceee0ea0-195d-4c58-8b57-4013c81e373c","title":"Member of Technical Staff - Generalist","department":"Member of Technical Staff","team":"Member of Technical Staff","employmentType":"Intern","location":"San Francisco","secondaryLocations":[],"publishedAt":"2026-11-05T19:13:57.308+00:00","isListed":true,"isRemote":false,"workplaceType":"OnSite","address":{"postalAddress":{"addressRegion":"California","addressCountry":"United States","addressLocality":"San Francisco"}},"jobUrl":"https://jobs.ashbyhq.com/trajectory/ceee0ea0-195d-4c58-8b57-4013c81e373c","applyUrl":"https://jobs.ashbyhq.com/trajectory/ceee0ea0-195d-4c58-8b57-4013c81e373c/application","descriptionHtml":"As a Generalist Engineer at Trajectory, you will craft the product experience that makes continual learning legible and intuitive.
You will own the SDK, API surfaces, and integration infrastructure that help customers understand, debug, and trust continually improving models. This includes trajectory visualization, evaluation workflows, monitoring dashboards, and the core product interfaces that sit on top of our leading RL infrastructure.
You’ll work directly with the founders and industry-leading design partners (Series B+ AI vertical product companies) to translate frontier RL systems into a clean, developer-first product experience. You will help define the visual language, interaction patterns, and product philosophy of Trajectory.
We are building the interface for continual learning. This role sets that bar.
Strong full-stack or backend engineering experience
Experience building AI-powered products or agent systems - you actively experiment with new frameworks for fun
Experience building developer-facing products or SDKs
Experience with data pipelines, telemetry systems, or event infrastructure
Ability to move quickly in a zero-to-one environment
Experience building observability, tracing, or monitoring systems
Experience at early-stage developer tooling startups
Strong product instincts and ability to work directly with customers
Experience owning SDK design and developer ergonomics
Trajectory is a research and product lab creating the platform for continual learning.
AI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.
Trajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.
Founded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox.
","descriptionPlain":"JOB DESCRIPTION\n\nAs a Generalist Engineer at Trajectory, you will craft the product experience that makes continual learning legible and intuitive.\n\nYou will own the SDK, API surfaces, and integration infrastructure that help customers understand, debug, and trust continually improving models. This includes trajectory visualization, evaluation workflows, monitoring dashboards, and the core product interfaces that sit on top of our leading RL infrastructure.\n\nYou’ll work directly with the founders and industry-leading design partners (Series B+ AI vertical product companies) to translate frontier RL systems into a clean, developer-first product experience. You will help define the visual language, interaction patterns, and product philosophy of Trajectory.\n\nWe are building the interface for continual learning. This role sets that bar.\n\n\n\n\n\nREQUIRED QUALIFICATIONS\n\n - Strong full-stack or backend engineering experience\n\n - Experience building AI-powered products or agent systems - you actively experiment with new frameworks for fun\n\n - Experience building developer-facing products or SDKs\n\n - Experience with data pipelines, telemetry systems, or event infrastructure\n\n - Ability to move quickly in a zero-to-one environment\n\n\nPREFERRED\n\n - Experience building observability, tracing, or monitoring systems\n\n - Experience at early-stage developer tooling startups\n\n - Strong product instincts and ability to work directly with customers\n\n - Experience owning SDK design and developer ergonomics\n\n\n\n\nABOUT TRAJECTORY\n\nTrajectory is a research and product lab creating the platform for continual learning.\n\nAI is the most capable software ever built, and the least able to learn. Every valuable correction and edit that happens in a product evaporates at the next session. A few teams have closed this gap by hand-coupling their models to their products: Composer, Claude Code, Windsurf SWE-1.\n\nTrajectory is the first scalable approach for every company: our platform unlocks the signal already sitting in product use, so companies can continuously post-train large-scale agentic models that outperform the frontier. Working with leading AI companies, including Decagon, Clay, Harvey, Mercor, and Rogo.\n\nFounded by researchers from Deepmind, OpenAI, Meta, Apple, Amazon, and Scale. $15M led by Conviction, with Fei-Fei Li, Jeff Dean, and founders of Notion, Braintrust, Modal, Hugging Face, and Dropbox."}],"apiVersion":"1"}