Generative AI Engineer

Location

Years of Experience - 2-5 Years

India · Hybrid (Mumbai / Bengaluru / Chennai / Hyderabad) · 20–30% client-site travel

Apply Now - Click on apply now

About the Role

You've shipped real software. You've sat in a room with stakeholders and pushed back on a bad idea. You've built something with LLMs that ran in production and learned, the hard way, what doesn't make it past the demo. Now you want to do more of that, faster, with bigger problems and better tools.

As a mid-level FDE at Firstsource, you'll be embedded with a named client, own a problem end-to-end, and partner with junior FDEs and our internal AI platform team to build solutions that operate — not just impress. You'll be expected to think about extensibility on day one and ship working prototypes by day five.

What You'll Own

• End-to-end delivery of Gen AI solutions inside a client engagement — from discovery to production
• Architecture decisions on document pipelines, agent systems, RAG infrastructure, copilots, and operational AI
• Hardening LLM features: evals, prompt versioning, cost & latency optimisation, observability, guardrails
• Stakeholder management — translating client problems into technical scope, and technical realities back into business language
• Mentoring 0–2 year FDEs on the engagement, reviewing their code, and accelerating their ramp
• Contributing patterns and platform components back to Firstsource's reusable AI stack

What We're Looking For

• 2–5 years of engineering experience with at least one product/major feature shipped to real users (can speak to scale, edge cases, what broke)
• Strong Python + one of TypeScript, Go, or Java; comfortable across the stack when needed
• Production Gen AI experience — you've built and deployed an LLM-powered feature, not just a demo. You understand cost, latency, prompt versioning, and have set up evals
• Solid system design — distributed systems concepts, vector DBs (Pinecone, Weaviate, pgvector), cloud (AWS / Azure / GCP)
• Direct experience with stakeholders, PMs, or clients — not buried inside a pure backend pod
• At least one 0 to 1 project you owned end-to-end
• Fluent with Gen AI pair programming (Claude Code, Cursor, Copilot, Windsurf) — uses them as a force multiplier, reviews output critically

Stretch / Strongly Preferred

• Built or contributed to internal tooling/platforms used by other engineers
• Vertical domain depth — BFSI, healthcare, telecom, retail, or insurance
• Eval frameworks experience (Braintrust, LangSmith, internal eval pipelines)
• Experience with MCP servers, agent frameworks, multi-agent orchestration
• Have broken production at least once on an architectural choice — and can articulate the lesson

What We Offer

• Named, high-stakes client engagements with real budgets and real business outcomes
• A path to Senior / Staff FDE and architect roles within 2–3 years for clear performers
• Latest Gen AI tooling and platform access — Claude Code, internal AI platform, vector DBs, eval pipelines
• Compensation aligned to outcomes — top FDEs are compensated like senior engineers, not by tenure tables
• Direct exposure to Firstsource leadership and the wider RP-Sanjiv Goenka Group network