RAG / LLM Engineer

We are seeking a skilled RAG / LLM Engineer to develop production-ready Retrieval-Augmented Generation (RAG) systems, prioritizing accuracy, performance, and scalability. This role involves managing enterprise documents and optimizing LLM outputs for practical applications in Chennai.

Roles & Responsibilities

• Construct end-to-end RAG pipelines encompassing ingestion, chunking, embeddings, retrieval, and generation.
• Implement hybrid search strategies, combining vector search with BM25 or keyword-based methods.
• Optimize document chunking for both structured and unstructured data.
• Work with a variety of Large Language Models (LLMs) such as LLaMA, Qwen, Mistral, Gemma, and APIs from OpenAI/Claude.
• Conduct prompt engineering for question-answering, summarization, and data extraction tasks.
• Mitigate hallucinations by employing guardrails and fine-tuning retrieval mechanisms.
• Utilize popular vector databases including FAISS, Chroma, Pinecone, and Milvus.
• Implement advanced search functionalities like semantic search, re-ranking, and query expansion.
• Process diverse document formats such as PDFs, DOCX, Excel, and scanned documents.
• Integrate Optical Character Recognition (OCR) technologies like Tesseract and PaddleOCR.
• Optimize for low latency and high throughput, incorporating caching strategies.
• Build robust APIs using FastAPI.
• Deploy solutions on AWS, GCP, or on-premise environments using Docker.

Required Skills

• Proficiency in Python programming.
• Demonstrated experience with LLMs and RAG systems.
• Solid understanding of Natural Language Processing (NLP), embeddings, and vector search.
• Familiarity with LangChain / LlamaIndex frameworks.
• Experience with vector databases (FAISS, Pinecone, Milvus).
• Proficiency in FastAPI / Flask for API development.
• Basic knowledge of Git and CI/CD practices.

Good to Have Skills

• Experience with hybrid search (BM25 + vector).
• Knowledge of re-ranking models.
• Familiarity with OCR and layout-aware models (e.g., Donut, LayoutLM).
• Experience with Ollama / llama.cpp.
• Understanding of GPU/CPU optimization techniques.
• Experience with offline or air-gapped deployment scenarios.

Experience Requirements

• 2–4 years in Backend, NLP, or Machine Learning roles.
• Minimum of 1 year of hands-on experience with LLMs or RAG systems.