Job description
Product Engineer | Fibr.
ai Bengaluru · In office · 2-5 years Stack: Python · FastAPI · Temporal · PostgreSQL · MongoDB Fibr is building the agentic web experience layer - turning every URL into an intelligent agent that senses intent, makes decisions, and adapts in real time.
Our agents personalise the full journey, running long LLM workflows against live customer accounts and millions of sessions.
You'll own features end-to-end - from the rough idea in a Slack thread to something shipped, instrumented, and behaving correctly in production.
New surface areas land on your plate every few weeks: a new agent, a new integration, a new optimization loop.
You're responsible for the whole thing, including the evals and metrics that prove it works.
You'll work directly with the founding team and ship to real users every week.
What you'll do Ship AI-powered features end-to-end - backend, data, LLM layer, and the surface the user sees Design long-running workflows that hold up against rate limits, partial failures, and noisy third-party APIs Build and maintain the evals, guardrails, and instrumentation that decide whether a feature is good enough to ship Model the data behind agents and the analytics behind product decisions Sit in on architecture calls and shape where the platform goes next Nice-to-haves RAG, vector stores, or tool-use / multi-step agents in production Experience with rate-limited third-party APIs at scale (ad platforms, CRMs, analytics) Strong product sense - you catch the UX issue before the user does Time spent at an early-stage startup How we work High ownership, low bureaucracy.
Decisions in the room, not in docs.
You'll work directly with the founding team.
Competitive salary + meaningful early-stage equity A chance to build deep expertise at the frontier of AI product development Private Health Insurance Performance Bonus 2-5 years of production backend experience Has shipped LLM-powered features to real users in production - not demos, not side projects.
Be ready to talk about what broke and how you fixed it Has built and run evals for LLM systems.
Show us the harness, the dataset, and the decisions it drove Comfortable across relational and document databases Has worked with a durable workflow or queue system in production Ships fast without breaking the build
This job post has been translated by AI and may contain minor differences or errors.
