AI / ML Engineer Resume Sample for India (2026)
A real ATS-friendly resume sample for AI / ML Engineer (2-5 years, AI / Machine Learning).
Sample resume
Ananya Krishnan — Bengaluru, India
Applied AI engineer (3.5 yrs) shipping GenAI products in production. Built RAG + agent systems serving 50K+ daily queries at a Bangalore fintech. Strong on Python, LLM application patterns, vector search, and evals. Transitioned from backend engineering into AI in 2024. Looking for a Senior AI Engineer role where I own GenAI features end-to-end.
Experience
AI Engineer — Setu (Bangalore, hybrid) (Jun 2024 – Present)
- Built a production RAG system over 2M+ financial documents (FastAPI + pgvector + GPT-4o); answers 50K+ daily support queries with 91% answer-accuracy on the eval set.
- Cut LLM inference cost 58% by adding semantic caching (Redis) + routing simple queries to a smaller model — saved ~₹6L/month at scale.
- Designed the eval framework (200+ labelled cases) that gated every prompt change; eliminated 3 regressions that would have shipped to production.
- Built an agent with tool-use that automates merchant-onboarding doc verification — reduced manual review time 40%.
Backend Engineer — Razorpay (Bangalore) (Jul 2021 – Jun 2024)
- Owned the settlement-reconciliation service (Python/PostgreSQL) processing 3M+ daily transactions at 99.96% uptime.
- Self-initiated the team's first ML use case (fraud-signal scoring) — which became my entry into AI engineering.
Education
B.Tech, Computer Science Engineering, NIT Warangal (2017 – 2021) — CGPA: 8.7 / 10.0
Skills
- Languages: Python, SQL, TypeScript
- AI / LLM: RAG, Agents / tool-use, Prompt engineering, Evals, Fine-tuning (LoRA), OpenAI / Anthropic / Gemini APIs
- ML / Data: PyTorch (applied), scikit-learn, pandas, Hugging Face
- Vector / Infra: pgvector, Pinecone, Redis, Docker, AWS (EC2, S3, Bedrock), LangChain / LlamaIndex
Projects
docchat — Open-source RAG-over-PDF library
- Python library for production RAG over PDFs with smart chunking + hybrid retrieval; 640+ GitHub stars, used in 4 production codebases.
- Includes a built-in eval harness — the feature most users cite, because most RAG libraries skip evaluation entirely.
Hindi-English code-mixed sentiment model
- Fine-tuned a small LLM (LoRA) for code-mixed Hinglish sentiment, common in Indian customer support; 84% F1 on held-out data.
- Published model + dataset on Hugging Face; 2K+ downloads.
Why this resume works for ATS
- Leads with AI work, not the backend past. The summary + first role are AI-first — a recruiter filtering for "AI Engineer" sees the match in 3 seconds. The backend role is present but secondary, framed as the on-ramp into AI.
- Every bullet has a number AND a system detail (50K queries, 91% accuracy, 58% cost cut, ₹6L/month saved). AI hiring managers are flooded with "prompt enthusiasts" — quantified production impact separates real engineers instantly.
- Shows eval discipline — "designed the eval framework" signals senior AI thinking. Most candidates can prompt; few can measure. This is the single most differentiating line on the resume.
- Cost-awareness is called out ("cut inference cost 58%"). AI features die on unit economics; engineers who optimize cost are disproportionately valued in 2026.
- Open-source RAG library + Hugging Face model = demonstrated capability beyond the day job. At this experience band, public AI work carries enormous weight.
Common mistakes for AI / ML Engineer resumes
- Listing "ChatGPT" or "Prompt Engineering" as your only AI skill. That signals hobbyist. Show the full applied stack — RAG, agents, evals, vector DBs, deployment.
- No evals mentioned anywhere. This is the #1 tell of a prompt tinkerer vs a real AI engineer. If you've measured output quality systematically, say so prominently.
- Burying real AI projects under a wall of completed courses. One shipped RAG app outweighs ten Coursera certificates. Cut the certificate list to 1-2 credible ones.
- Claiming "ML researcher" skills (deriving architectures, novel research) when you do applied work. Mismatched framing gets caught in interviews. Be precise about Applied AI vs ML Engineer vs Research.
- Generic "worked on AI features" bullets with no scale or outcome. "Built a RAG system serving 50K daily queries at 91% accuracy" is hireable; "worked on a chatbot" is invisible.
Frequently asked
How do I show AI engineering experience if my day job is still backend?
Pull from side-of-desk AI work, self-initiated projects, and personal builds. The candidate above used a fraud-scoring ML use case she started at her backend job as the bridge. Build 1-2 real LLM projects (a RAG app, an agent) outside work, deploy them, open-source them, and lead your resume with those. Demonstrated AI capability beats a job title.
What AI skills should be on a 2026 AI engineer resume?
For Applied AI roles: Python, RAG, agents/tool-use, prompt engineering, evals, vector databases (pgvector/Pinecone), one LLM framework (LangChain/LlamaIndex), and deployment + cost optimization. Eval ability is the most differentiating — most candidates can prompt but few can measure output quality systematically.
Should I list every AI/ML course I have completed?
No. Cut to 1-2 credible ones at most. AI hiring managers heavily discount certificate lists — they signal "studied" not "shipped." One deployed AI project (with a GitHub link and an eval harness) outperforms ten course completions on an AI engineering resume.
Do I need a PhD to get an AI engineering job in India?
No — not for Applied AI / GenAI Engineer roles, which are ~70% of current demand. These reward applied LLM engineering (RAG, agents, evals, deployment) plus strong software skills. PhDs are typically needed only for ML Research / Research Scientist roles. Match your resume framing to the specific track.