Senior Software Engineer

About the Company
At our core, we are an AI-powered entertainment company with a deeply human-first philosophy. We believe technology should amplify creativity. Our proprietary AI systems work alongside writers, voice artists, and creative teams to help stories scale globally, faster, smarter, and with cultural depth and emotional integrity intact.
Today, Pocket FM is home to a vibrant community of 250+ million listeners, 300,000+ creators, and 100,00+ audio series. With over 140 billion minutes streamed annually, we have emerged as one of the fastest-growing media-tech companies in the world, and we are just getting started. We operate at a massive global scale, but with a startup mindset: curious, fast-moving, and deeply owner-driven.
At Pocket FM, teams are encouraged to think boldly, move with intent, and build for long-term impact as we shape the future of audio-first storytelling worldwide
Our unique model combines free listening with micropayments for premium content, powering strong business growth. In FY25, we reached an ARR of INR 2,000 crore, with over 100,000 hours of content on the platform. We're also at the forefront of innovation, leveraging AI-generated content to scale efficiently.
About the Role
AI Engineer 2 — Agentic Systems & Retrieval
Responsibilities
Build and own agentic systems: Agent loops, tool calling, routing, memory, guardrails, state management, and streaming using modern agentic frameworks.
Build production RAG systems: Own the full retrieval pipeline — chunking, embeddings, hybrid/vector search, reranking, graph retrieval, and context packing.
Own prompting & context engineering: Design and optimize prompts, context strategies, tool instructions, and memory for quality, reliability, and efficiency.
Work deeply with LLM models: Evaluate and route across models/providers based on quality, latency, reliability, and cost.
Own AI quality & evals: Build golden sets, retrieval/generation evaluations, regression suites, and production quality metrics.
Own systems end to end: Take AI capabilities from problem definition and design through implementation, testing, production rollout, monitoring, and iteration.
Own production reliability: Debug agent failures, retrieval misses, token spikes, latency issues, and other production problems; drive permanent fixes.
Build scalable AI services: Strong Python services with async APIs, databases, background jobs, observability, and production-grade testing.
Raise the engineering bar: Drive design/code reviews, documentation, knowledge sharing, and effective use of AI coding/debugging tools.
Required Skills
4+ years of software engineering experience, including 2+ years shipping production LLM applications.
Strong Python with production experience in async APIs, SSE, testing, and distributed/background systems.
Hands-on experience with agentic frameworks, LLM models, prompting, context engineering, and production RAG systems.
Experience owning a production agent or multi-step tool-using system end to end.
Experience building and operating RAG systems, including measuring and fixing retrieval failures.
Hands-on experience with relational, vector, and graph databases in production.
Experience building or operating an LLM/AI eval framework that catches regressions before users do.
Strong understanding of quality, latency, reliability, and token-cost trade-offs, backed by data.
Strong ownership and ability to independently take ambiguous AI problems from definition → production → measurable outcome.

Work arrangement
Hybrid