AI Lead

BDO India; Mumbai, Navi Mumbai; On-site; INR 1,800,000 - 2,700,000; Full time

Job summary

BDO India seeks a skilled AI/ML engineer to implement robust RAG pipelines for document ingestion, embedding and retrieval. You will design end-to-end LLM-driven workflows with tool use, loops, and error handling, and integrate LLM APIs into product features for extraction, classification and summarisation.

You will also develop FastAPI endpoints, contribute to architectural decisions, and maintain an automated evaluation suite to ensure regression-free performance across AI features.

Qualifications

• Python—clean, reviewable, testable code; not script-level, production-ready.
• LLM API usage with OpenAI/Anthropic/Azure; streaming, function calling, token management.
• RAG end-to-end: retrieval pipeline, embeddings, vector store loading, and integration.

Responsibilities

• Implement RAG pipelines for document ingestion, chunking and embedding.
• Build agentic workflows with LLM-driven processes, tool use, loops and error handling.
• Integrate LLM APIs into product features for extraction, classification and summarisation.
• Write and maintain an LLM evaluation suite with test cases and regression tracking.
• Build FastAPI endpoints to expose AI capabilities to services and front-ends.
• Participate in design reviews and contribute to architecture decisions.

Skills

Python

LLM API usage

RAG implementation

Vector databases

FastAPI

Git & Docker

Tools

LangChain

LLM evaluation

Agent design

Fine-tuning

Job description

RESPONSIBILITIES

• Implement RAG pipelines document ingestion, chunking, embedding, vector store loading, retrieval and response generation
• Build agentic workflows — LLM-driven processes with tool use, loops and error handling
• Integrate LLM APIs into product features — extract, classify, summarise and generate structured outputs
• Write and maintain an LLM evaluation suite — test cases, metrics, regression tracking
• Build FastAPI endpoints that expose AI capabilities to other services and front-ends
• Participate in design reviews and contribute to architecture decisions

REQUIRED SKILLS

• Python — clean, reviewable, testable code. Not script-level, production-level.
• LLM API usage — OpenAI, Anthropic or Azure OpenAI SDK. Streaming, function calling, token management.
• RAG implementation — has built at least one retrieval pipeline end-to-end
• Vector databases — has loaded and queried at least one (Pinecone, Chroma, Weaviate, pgvector)
• FastAPI — has built and deployed at least one REST API
• Git and Docker — standard development workflow

ADDITIONAL SKILLS (increase placement level)

• LangChain, LlamaIndex or LangGraph X— used in a real project, not documentation examples
• LLM evaluation — has built automated test suites, not just manual review
• Agent design — can design a tool-use loop with memory and error recovery, not just implement one
• Fine-tuning — has fine-tuned a model (LoRA/QLoRA) and understands when to use it