Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
An Agent Engineer owns a client's voice or chat agent in production. Not a component of it, the whole thing: the procedure it follows, the instructions that govern how it speaks, the code that connects it to the client's systems, and the tests that prove it does what the documentation says. When a client says \"the bot offered a payment plan below our minimum,\" you're the person who works out why, fixes it, and explains the fix to the client. You'll ship changes that speak to real callers in your first two weeks.
Own a client delivery end to end. Take requirements from first conversation through pilot, production, and continuous iteration as procedures change, volumes grow, and models improve.
Engineer the agent's behavior. Write and maintain the instruction sets that determine what the agent says and when. Instructions are versioned, tested, and reviewed like code, because a wrong word in a disclosure is a compliance incident, not a UX bug.
Build the evaluation gate. Automated tests that replay past conversations and check the agent follows the procedure. Nothing ships without passing them.
Debug live conversations. Look up what the system knew at each moment, read the exact instructions the model was given, and work out why it responded the way it did.
Work directly with clients. Sit in on calls with US enterprise clients, own the technical conversation, and watch your changes move their business metrics. Not many engineering roles put you this close to the people using what you build.
Recent work by engineers on this team, all on live debt collection agents:
Build an intent identification layer for a voice agent, so every response is grounded in what the caller actually asked rather than what the model assumes, sharply reducing hallucinations on live calls
Extend the negotiation flow so the agent offers payment plans only within the limits the client has set, with guardrails that make out-of-bounds offers impossible rather than just unlikely
Grow the eval suite that replays real collection calls and verifies every legally required disclosure was delivered, word for word, before any release ships
Python with FastAPI, PostgreSQL, Temporal for background workflows, WebSockets holding real-time conversations open, and STT and TTS providers on the speech side, with LLMs via OpenAI and similar underneath.
Everything runs on Kubernetes with ArgoCD-driven GitOps deployments, SigNoz for observability, DeepEval driving the eval suite our CI runs before anything reaches production, Ory and OpenFGA for identity and access management, and a lot of beautifully built internal agent architecture underneath.
2 to 5 years building and running production systems. You've been on call for something you built and debugged it under pressure.
You write Python another person can read, review others' work thoughtfully, and debug by forming a theory and testing it, not by changing things until they work. Reading code you didn't write and working out what it does is the single most important skill in the role.
You've worked with async code and a relational database.
You're comfortable in front of a client: take vague feedback, ask the right questions, leave with a scoped change.
Strong written communication. Agent instructions, client explanations, and incident writeups are all writing, and an agent is only as precise as the instructions behind it.
LLM systems in production: eval frameworks, agent tooling, RAG pipelines, structured prompting
Conversational AI experience, voice or chat: dialogue design, IVR systems, chatbots, speech interfaces
A regulated industry: finance, healthcare, insurance, collections
Real-time or telephony systems: WebSockets, streaming audio, Twilio or similar
Founder or founding engineer experience
Explicitly not required: prior \"AI engineering\" as a job title, or a computer science degree. Careful engineers who own outcomes pick this up fast.
Small teams per client, with real ownership. You'll have peers to pair with and review your work, and you'll review theirs. We review all code, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
While Agent Engineers own individual client deliveries, you own the platform underneath all of them: the runtime that agents execute on, the voice infrastructure that keeps live calls fast and stable, the enterprise capabilities that get us through security reviews, and the intelligence layer that turns millions of conversations into insight. This is the highest-leverage engineering seat in the company: every improvement you ship reaches every client at once.
You'll work directly with the founder on architecture and own your decisions end to end, from design through production and the incident channel.
Build the agent platform primitives. Design and evolve the runtime, orchestration engine, and the systems behind Agent Studio: how agents are defined from written procedures, how they reason and take actions, and how guardrails make out-of-bounds behavior impossible rather than unlikely. Building agents is its own engineering discipline, and you'll be building the tooling that discipline needs.
Own the voice infrastructure. Real-time audio streaming over WebSockets, STT and TTS integration at low latency, and the hard parts of live conversation: interruptions, disconnects, transfers, and telephony integrations that pick up and place calls reliably at growing volume.
Form the foundation of enterprise trust. Clients hand us regulated conversations with their own customers. You'll build and uphold the capabilities that make that possible: RBAC, access management, audit logs, data isolation, and integrations that embed into complex client environments while surviving their security reviews.
Build the evaluation layer. We believe evaluation methodology, not model capability, is the bottleneck, and you'll own the system that proves it: the DeepEval-driven suite that replays real conversations against every change, simulation of the scenarios we haven't seen yet, and experimentation frameworks that let us A/B test agent behavior and make changes with evidence instead of intuition. Every release gates on what you build here.
Build the observability layer. Observability goes beyond logging. You'll extend Reflect so that every agent decision is traceable: what the system knew at each moment, which instructions the model was given, and why it responded the way it did. Add proactive monitoring that surfaces regressions, drift, and new conversation patterns before a client notices, and turn millions of calls into insight the whole company acts on.
Close the feedback loop. Connect what evaluation and observability find back into how agents improve: recurring failure modes become new eval cases, production patterns reshape procedures, and agents get measurably better over time.
Keep the platform fast, reliable, and boring to operate. Own Kubernetes, ArgoCD-driven GitOps deployments, CI/CD, and observability with SigNoz. Build the self-serve infrastructure that lets the rest of engineering ship without waiting on you, and lead incident response and postmortems when things break.
Python with FastAPI, PostgreSQL, Temporal for background workflows, WebSockets holding real-time conversations open, and STT and TTS providers on the speech side, with LLMs via OpenAI and similar underneath.
Everything runs on Kubernetes with ArgoCD-driven GitOps deployments, SigNoz for observability, DeepEval driving the eval suite our CI runs before anything reaches production, Ory and OpenFGA for identity and access management, and a lot of beautifully built internal agent architecture underneath.
5+ years of hands-on experience building and operating production systems, with strong backend and distributed systems fundamentals.
Proven experience with cloud platforms, infrastructure as code, and container orchestration: you've run Kubernetes in production, not just in a tutorial.
Real-time systems depth: you understand latency budgets, streaming, backpressure, and what makes a live audio conversation different from a request-response API.
Experience with observability tooling (SigNoz, OpenTelemetry, or similar) and with incident response: you've been paged, found root cause, and made the pager quieter afterward.
You care that a system stays reliable, secure and maintainable as it grows, and the judgment to know when boring technology is the right answer.
High agency: you drive outcomes in a high-autonomy environment, find creative ways around obstacles, and don't wait to be told what matters.
Degree in computer science or a related field, or equivalent professional experience.
Experience building enterprise features: SSO, RBAC, IAM, audit logs, data isolation, or compliance-adjacent systems
Production experience with LLMs, agent frameworks, retrieval, or evaluation systems
Telephony or streaming audio experience: Twilio, WebRTC, SIP, or contact center integrations
Experience with large-scale data systems, analytics platforms, or ML-powered product features
Experience building developer platforms, SDKs, or internal tooling other engineers love
Leadership experience on technical projects or teams
Small teams, real ownership. You'll have peers to pair with and review your work, and you'll review theirs. We review all code, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
Sales at Rifa is founder-led today. As GTM Lead, you take that motion and turn it into a machine: you'll own the pipeline, close enterprise deals yourself, and build the playbook that the sales team we hire after you will run. You'll report to and work directly with the founder on every strategic decision about how Rifa goes to market.
Carry the number. You own pipeline and closed revenue for the first non-founder sales motion at Rifa. Founder-led deals won us paying enterprise clients; your job is to make that repeatable without the founder in every room.
Sell to the people who can say no. An operations leader wants the agent. A compliance officer decides whether it ships. You'll run both conversations, and you'll get good at the third one: the vendor security review that determines how fast a signed deal becomes a live deployment.
Write the playbook as you go. Which accounts are worth a call, which objections keep recurring, what evidence moves a pilot into production. The sellers we hire after you will run what you wrote, so write it down as you learn it.
Scope pilots that convert. Work with agent engineers to define a pilot on the client's own call volume, with success criteria a compliance team will actually sign off on. A pilot nobody can measure is a pilot that never closes.
Feed the market back into the product. You'll hear the same three objections before anyone else does. Bring them to the founder and to engineering while they're still cheap to fix.
Sell the way our product works. We build agents that do real work. A seller doing account research and follow-up by hand is a poor advert for that, so automate the low-value half of your week and spend it on customers instead.
Six to eight years selling enterprise software, closing six- and seven-figure ACV deals you sourced yourself rather than inherited.
Experience selling to a regulated buyer, in financial services, collections, insurance or healthcare. You treat \"legal and compliance are reviewing it\" as a stage to be moved, not a stall to wait out.
You've been early somewhere. No SDR feeding you, no marketing team producing collateral, and you built pipeline anyway.
Enough technical fluency to hold the room without an engineer in it, and the judgement to know where that ends. Our buyers ask how an agent is tested and what happens the first time it gets something wrong.
Clear writing. A cold email, a pilot scope and a mutual action plan are all writing, and on a team this size nobody rewrites it for you.
You've sold into collections, lending or accounts-receivable operations, and you know what a BPO contract looks like from the inside.
You've sold conversational AI, contact centre software or CCaaS, and can tell a buyer honestly where the category over-promises.
You've been a first sales hire before and have opinions about what you'd do differently.
Comfort with slow procurement: security questionnaires, pen-test reports, DPAs, and the patience they demand.
Small team, real ownership, founder-led deals you'll increasingly run yourself. We review each other's work, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
Rifa ships voice and chat agent releases to multiple enterprise clients on overlapping timelines, and every one of those releases has an eval gate, a client sign-off, and a rollout plan. You'll own that machinery, reporting directly to the CEO. As Technical Program Manager you run the release train end to end: planning what lands in each cycle, tracking dependencies across engineering, QA, and client-facing teams, catching risk before it becomes a slipped date, and making sure a release that passes internally actually lands cleanly with the client.
This is a role for someone who reads engineering tickets and QA reports as fluently as a project plan, and who keeps five threads moving without letting any of them become someone else's surprise.
Run the release train. Own the release calendar across voicebot and chatbot versions and across clients. Plan each cycle's scope with engineering, hold the eval gate as a hard line, and manage staged rollouts so no client is surprised by what changes.
Track dependencies and kill risk early. Map what blocks what across engineering, QA, infrastructure, and client schedules. Surface slippage when it's a one-day problem, not a one-week one, and drive the trade-off conversation to a decision.
Own program communication. Keep a single, current source of truth for every program: status, risks, decisions, and owners, written clearly enough that the CEO, an engineer, and a client-facing teammate all act from the same picture.
Run client onboarding and migration programs. When a new client goes live or an existing one moves to a new agent version, you own the plan: milestones, environment readiness, data and integration checkpoints, sign-offs, and the go-live itself.
Improve the machinery. Where the process creates friction, fix the process. Where a checklist would have caught the incident, write the checklist. Keep it simple; process is a cost we pay only when it clearly buys something.
4 to 7 years in technical program management, release management, or engineering-adjacent project management at a B2B software company, with at least part of that in a production environment where releases had real consequences.
Enough technical depth to be dangerous: you can read an engineering ticket, follow an architecture discussion, and understand why a failing eval blocks a release, without needing an engineer to translate.
A track record of shipping cross-team programs on time, and of being honest early when they won't be.
US-hours overlap: reliable daily overlap with US business hours (ET/CT), which in practice means a late-afternoon-into-evening IST schedule, since client rollouts and go-lives happen on their clock.
Strong written and spoken English. Status updates, rollout plans, and risk callouts are the core artifacts of this job, and they need to hold up under scrutiny.
Experience shipping AI/ML or conversational AI products, or working with eval-gated release processes
Experience in a regulated domain: financial services, collections, healthcare, insurance
Familiarity with observability and QA tooling, or with client-facing rollout and migration programs
Early-stage startup experience where you built the process rather than inherited it
Small teams per client, with real ownership. We review each other's work, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
This is the hands-on quality and operations role behind our enterprise client relationships. You'll live in the transcripts and performance data: auditing how our AI agents handle real client conversations, catching issues before and after clients do, and turning what you find into clear, actionable work for the engineering team.
You'll be client-facing: on audit calls, walking client-side QA and compliance teams through findings, and reconciling our read of bot performance with theirs. It's a role for someone precise, curious, and organized, comfortable with data and detail, and confident explaining what they found and why it matters.
You'll report to the Technical Program / Product Manager, who owns the overall client relationship, and you'll be their eyes and ears on day-to-day quality.
Audit transcripts and bot performance. Review call and chat transcripts in Reflect, our observability platform, evaluate how our AI agents performed against quality and compliance standards, and score conversations consistently and fairly.
Find and document issues. Spot regressions, compliance gaps, mishandled flows, and edge cases in agent behavior, and document them clearly, with the transcript or metric as evidence.
Convey issues to engineering. Translate what you find into well-written Linear cards that engineers can act on without a follow-up call: clear repro steps, the affected conversation, and why it matters to the client.
Get on client calls. Join client QA and audit calls, walk their teams through findings, and answer questions on how our agents are performing.
Reconcile audits with the client. Compare our QA findings against the client's own audit team's results, work through discrepancies, and align on a shared, defensible view of quality.
Support the client relationship. Feed insights, recurring issues, and trends back to the Technical Program / Product Manager to inform release communications, weekly syncs, and monthly reviews.
1 to 3 years in QA, operations, quality analysis, delivery support, or a similar client-facing role, ideally at a B2B SaaS, BPO, or financial-services company.
Comfort with data and transcripts as evidence. You can read a call log or a QA scorecard, form your own view, and back it up, not just relay what someone else said.
Clear written and spoken English. A large share of this job is written: issue tickets, audit notes, and client-facing summaries that hold up under scrutiny.
US-hours overlap. Reliable daily overlap with US business hours (ET/CT), which in practice means a late-afternoon-into-evening IST schedule, and comfort being on a call with a client's ops or compliance team.
Precision and follow-through. You're organized, detail-oriented, and you close the loop on what you find.
Experience in debt collections, lending, or another regulated financial-services domain, or familiarity with FDCPA / compliance workflows
Hands-on experience with QA tooling, call-quality scoring, or observability/audit dashboards
Familiarity with Linear (or similar issue trackers like Jira) and how good engineering tickets are written
An interest in AI / conversational systems and how they behave in production
Small teams per client, with real ownership. We review each other's work, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
Our agents phone people about overdue accounts. It is not a friendly context, and the voice carries more of the load than anyone expects: it has to sound like a person, hold steady when the other side gets angry, and read a required disclosure without sounding like a recording of a lawyer.
Nobody here owns that today. Engineers pick whichever voice the TTS provider offers and move on, and you can hear it on the calls. We want someone whose actual job is how Rifa sounds in US English.
Most of your first few months goes on US English voices for live collections calls. The rest goes on making the work repeatable, so the next locale is easier than this one.
Start with the compliance language. Every call carries wording that has to come out verbatim. Engineering guarantees the words are right. You decide whether they land as information or as a threat, and that decision shows up in whether people stay on the line. Most candidates underrate this part of the job.
Design the voices. Accent, pace, where the stress falls, how long a pause can run before it reads as hesitation. These are judgement calls, and you will have to defend them to a client who asks for \"warmer\" and cannot say what warmer means.
Work in the pipeline. Voice config lives in the repo. You will edit it, run it, listen, adjust, open a PR. We are not building you a GUI for this, so you will need to be comfortable in a terminal. When you hit the limits of the tooling, say so precisely; vague complaints do not get fixed.
Cast and record. Find voice actors, direct the session, then do the unglamorous half: cutting, cleaning and prepping takes until the pipeline can use them.
Give us a way to argue about quality with evidence. Today \"does this sound good\" gets settled by whoever feels strongest about it. Naturalness, intelligibility down a bad phone line, whether a voice drifts over a twelve minute call: pick what actually matters and get it into the eval suite that already gates our releases.
Your ear and your portfolio matter more here than your job titles.
You have worked professionally on how things sound. Sound design, audio post, music production, game or interface audio: somewhere it counted.
You can record and edit speech yourself. Directing a nervous actor into a usable read, and salvaging a take from a room with a fridge in it.
A real ear for American English. Not just hearing that an accent exists, but being able to say why one read works on a collections call and another sounds like a telemarketer.
You are fine in a terminal. Scripts, config files, git, AI coding tools. Nobody expects you to write the synthesis code.
Something we can listen to.
Enough overlap with US afternoons to join client calls. In practice that means some evenings on IST.
TTS or voice cloning work: SSML, prosody controls, ElevenLabs and similar.
You know what 8 kHz narrowband does to a voice you spent a week on.
Somewhere tone had consequences: healthcare, financial services, a crisis line.
Another locale you would want to take on next. US English is only where we start.
Small teams per client, with real ownership. We review each other's work, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
An Agent Engineer owns a client's voice or chat agent in production. Not a component of it, the whole thing: the procedure it follows, the instructions that govern how it speaks, the code that connects it to the client's systems, and the tests that prove it does what the documentation says. When a client says \"the bot offered a payment plan below our minimum,\" you're the person who works out why, fixes it, and explains the fix to the client. You'll ship changes that speak to real callers in your first two weeks.
Own a client delivery end to end. Take requirements from first conversation through pilot, production, and continuous iteration as procedures change, volumes grow, and models improve.
Engineer the agent's behavior. Write and maintain the instruction sets that determine what the agent says and when. Instructions are versioned, tested, and reviewed like code, because a wrong word in a disclosure is a compliance incident, not a UX bug.
Build the evaluation gate. Automated tests that replay past conversations and check the agent follows the procedure. Nothing ships without passing them.
Debug live conversations. Look up what the system knew at each moment, read the exact instructions the model was given, and work out why it responded the way it did.
Work directly with clients. Sit in on calls with US enterprise clients, own the technical conversation, and watch your changes move their business metrics. Not many engineering roles put you this close to the people using what you build.
Recent work by engineers on this team, all on live debt collection agents:
Build an intent identification layer for a voice agent, so every response is grounded in what the caller actually asked rather than what the model assumes, sharply reducing hallucinations on live calls
Extend the negotiation flow so the agent offers payment plans only within the limits the client has set, with guardrails that make out-of-bounds offers impossible rather than just unlikely
Grow the eval suite that replays real collection calls and verifies every legally required disclosure was delivered, word for word, before any release ships
Python with FastAPI, PostgreSQL, Temporal for background workflows, WebSockets holding real-time conversations open, and STT and TTS providers on the speech side, with LLMs via OpenAI and similar underneath.
Everything runs on Kubernetes with ArgoCD-driven GitOps deployments, SigNoz for observability, DeepEval driving the eval suite our CI runs before anything reaches production, Ory and OpenFGA for identity and access management, and a lot of beautifully built internal agent architecture underneath.
2 to 5 years building and running production systems. You've been on call for something you built and debugged it under pressure.
You write Python another person can read, review others' work thoughtfully, and debug by forming a theory and testing it, not by changing things until they work. Reading code you didn't write and working out what it does is the single most important skill in the role.
You've worked with async code and a relational database.
You're comfortable in front of a client: take vague feedback, ask the right questions, leave with a scoped change.
Strong written communication. Agent instructions, client explanations, and incident writeups are all writing, and an agent is only as precise as the instructions behind it.
LLM systems in production: eval frameworks, agent tooling, RAG pipelines, structured prompting
Conversational AI experience, voice or chat: dialogue design, IVR systems, chatbots, speech interfaces
A regulated industry: finance, healthcare, insurance, collections
Real-time or telephony systems: WebSockets, streaming audio, Twilio or similar
Founder or founding engineer experience
Explicitly not required: prior \"AI engineering\" as a job title, or a computer science degree. Careful engineers who own outcomes pick this up fast.
Small teams per client, with real ownership. You'll have peers to pair with and review your work, and you'll review theirs. We review all code, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.
Rifa AI is building the AI agents platform for contact centers in regulated industries.
Enterprises in these industries want AI agents handling their customer operations and mostly can't deploy them. It's not a model problem. Horizontal platforms lack governance, release processes, and change management, and in a domain where every call can be reviewed by a regulator, that's disqualifying. Building an AI agent has never been easier. Deploying one an enterprise can trust has never been harder. That harder problem is the one we work on.
Our platform turns a company's written procedures into AI agents: voicebots that hold real-time conversations, take actions in the client's CRM, and stay within the limits the client has set. Every release is gated by an automated testing suite, and every conversation feeds a post-analysis platform. We're live in production today, handling debt collection calls for US financial services clients.
The engineering convictions behind it: evaluation methodology, not model capability, is the bottleneck. Every rule and decision trace becomes part of a company's context graph. Observability goes beyond logging. And our Agent Studio lets engineers, non-engineers, and auditors collaborate to build and improve AI agents with the right guardrails to achieve the expected business outcomes.
Rifa was founded by Sameer Fulzele (IIT Bombay). We're a small team of exceptionally capable, passionate engineers with paying enterprise clients and growing revenue, backed by Seaborne Capital, a founder-first firm of exceptional industry operators that works closely with its founders, and by angel investors who are veterans of enterprise software and operators in the accounts receivable industry.
While Agent Engineers own individual client deliveries, you own the platform underneath all of them: the runtime that agents execute on, the voice infrastructure that keeps live calls fast and stable, the enterprise capabilities that get us through security reviews, and the intelligence layer that turns millions of conversations into insight. This is the highest-leverage engineering seat in the company: every improvement you ship reaches every client at once.
You'll work directly with the founder on architecture and own your decisions end to end, from design through production and the incident channel.
Build the agent platform primitives. Design and evolve the runtime, orchestration engine, and the systems behind Agent Studio: how agents are defined from written procedures, how they reason and take actions, and how guardrails make out-of-bounds behavior impossible rather than unlikely. Building agents is its own engineering discipline, and you'll be building the tooling that discipline needs.
Own the voice infrastructure. Real-time audio streaming over WebSockets, STT and TTS integration at low latency, and the hard parts of live conversation: interruptions, disconnects, transfers, and telephony integrations that pick up and place calls reliably at growing volume.
Form the foundation of enterprise trust. Clients hand us regulated conversations with their own customers. You'll build and uphold the capabilities that make that possible: RBAC, access management, audit logs, data isolation, and integrations that embed into complex client environments while surviving their security reviews.
Build the evaluation layer. We believe evaluation methodology, not model capability, is the bottleneck, and you'll own the system that proves it: the DeepEval-driven suite that replays real conversations against every change, simulation of the scenarios we haven't seen yet, and experimentation frameworks that let us A/B test agent behavior and make changes with evidence instead of intuition. Every release gates on what you build here.
Build the observability layer. Observability goes beyond logging. You'll extend Reflect so that every agent decision is traceable: what the system knew at each moment, which instructions the model was given, and why it responded the way it did. Add proactive monitoring that surfaces regressions, drift, and new conversation patterns before a client notices, and turn millions of calls into insight the whole company acts on.
Close the feedback loop. Connect what evaluation and observability find back into how agents improve: recurring failure modes become new eval cases, production patterns reshape procedures, and agents get measurably better over time.
Keep the platform fast, reliable, and boring to operate. Own Kubernetes, ArgoCD-driven GitOps deployments, CI/CD, and observability with SigNoz. Build the self-serve infrastructure that lets the rest of engineering ship without waiting on you, and lead incident response and postmortems when things break.
Python with FastAPI, PostgreSQL, Temporal for background workflows, WebSockets holding real-time conversations open, and STT and TTS providers on the speech side, with LLMs via OpenAI and similar underneath.
Everything runs on Kubernetes with ArgoCD-driven GitOps deployments, SigNoz for observability, DeepEval driving the eval suite our CI runs before anything reaches production, Ory and OpenFGA for identity and access management, and a lot of beautifully built internal agent architecture underneath.
5+ years of hands-on experience building and operating production systems, with strong backend and distributed systems fundamentals.
Proven experience with cloud platforms, infrastructure as code, and container orchestration: you've run Kubernetes in production, not just in a tutorial.
Real-time systems depth: you understand latency budgets, streaming, backpressure, and what makes a live audio conversation different from a request-response API.
Experience with observability tooling (SigNoz, OpenTelemetry, or similar) and with incident response: you've been paged, found root cause, and made the pager quieter afterward.
You care that a system stays reliable, secure and maintainable as it grows, and the judgment to know when boring technology is the right answer.
High agency: you drive outcomes in a high-autonomy environment, find creative ways around obstacles, and don't wait to be told what matters.
Degree in computer science or a related field, or equivalent professional experience.
Experience building enterprise features: SSO, RBAC, IAM, audit logs, data isolation, or compliance-adjacent systems
Production experience with LLMs, agent frameworks, retrieval, or evaluation systems
Telephony or streaming audio experience: Twilio, WebRTC, SIP, or contact center integrations
Experience with large-scale data systems, analytics platforms, or ML-powered product features
Experience building developer platforms, SDKs, or internal tooling other engineers love
Leadership experience on technical projects or teams
Small teams, real ownership. You'll have peers to pair with and review your work, and you'll review theirs. We review all code, we write things down, and we'd rather hear an honest \"I don't know yet\" than a confident wrong answer.
Trust: We do the right thing, especially when nobody is checking. Clients hand us regulated conversations with their own customers, and we earn that every day.
Transparency: We write things down, share the real numbers, and say \"I don't know yet\" out loud, with each other and with clients.
Technically best solution: We choose what's right, not what's easiest or trendiest. When we notice we got it wrong, we fix it.
Decisiveness: We decide quickly with the information we have, commit, and correct course fast when reality disagrees.
Simplicity: We keep systems, processes, and words simple. Complexity is a cost we pay only when it clearly buys something.