RAG / LLM Engineer

Job Description

We are looking for a RAG / LLM Engineer to build production-ready Retrieval-Augmented Generation (RAG) systems with focus on accuracy, performance, and scalability. The role involves handling enterprise documents and optimizing LLM outputs for real-world use.

Roles & Responsibilities

• Build end-to-end RAG pipelines (ingestion, chunking, embeddings, retrieval, generation)
• Implement hybrid search (vector + BM25/keyword)
• Optimize chunking for structured and unstructured documents
• Work with LLMs (LLaMA, Qwen, Mistral, Gemma, OpenAI/Claude APIs)
• Perform prompt engineering for Q&A, summarization, extraction
• Reduce hallucinations using guardrails and retrieval tuning
• Use vector databases (FAISS, Chroma, Pinecone, Milvus)
• Implement semantic search, re-ranking, query expansion
• Process PDFs, DOCX, Excel, scanned documents
• Integrate OCR (Tesseract, PaddleOCR)
• Optimize latency and throughput, implement caching
• Build APIs using FastAPI
• Deploy on AWS/GCP/On-prem using Docker

Required Skills

• Strong Python programming
• Experience with LLMs and RAG systems
• Knowledge of NLP, embeddings, vector search
• LangChain / LlamaIndex
• Vector databases (FAISS, Pinecone, Milvus)
• FastAPI / Flask
• Git and CI/CD basics

Good to Have

• Hybrid search (BM25 + vector)
• Re-ranking models
• OCR and layout-aware models (Donut, LayoutLM)
• Ollama / llama.cpp
• GPU/CPU optimization
• Offline / air-gapped deployment experience

Experience

• 2–4 years in Backend / NLP / ML roles
• Minimum 1-year hands-on experience with LLMs or RAG systems