AI Engineer – Generative AI & RAG for Legal & Compliance

About the role

We are looking for an AI Engineer to design and deliver generative AI and agentic AI solutions that automate legal, regulatory and compliance document workflows. The role focuses on building high‑precision Retrieval‑Augmented Generation (RAG) pipelines and ensuring traceable, explainable AI outputs for regulated environments.

Key responsibilities

• Design and implement GenAI and agentic AI applications for complex document understanding, reasoning and decision support.
• Build and optimise RAG pipelines tailored to regulatory and legal documents, including contextual chunking, embeddings, metadata enrichment and semantic indexing.
• Develop reference identification, citation tracking and traceability mechanisms to guarantee auditability of AI outputs.
• Integrate Knowledge Graphs (RDF/SPARQL) with LLM workflows for structured and unstructured reasoning.
• Orchestrate multi‑step AI workflows using LangChain, LangGraph or similar frameworks.
• Establish AI quality‑assurance practices, including retrieval evaluation, hallucination detection and RAGAS‑style scoring.
• Build, train and fine‑tune specialised NER and document‑understanding models.
• Support end‑to‑end model lifecycle activities: experimentation, versioning, deployment readiness and monitoring.

Required profile

• 4–8 years of experience as an AI Engineer working on LLM‑powered solutions for legal, regulatory or compliance domains.
• Strong understanding of auditability, explainability and risk‑control requirements in AI systems.
• BE/B.Tech or equivalent engineering degree.

Required skills

• Python
• Docker / Docker Compose
• NLP / NLU
• GenAI / LLM application development
• Agentic AI
• Retrieval‑Augmented Generation (RAG)
• Embeddings and contextual chunking strategies
• Knowledge Graphs (RDF) and SPARQL
• Model lifecycle / ML application lifecycle
• LangChain, LangGraph
• Git
• AI QA / evaluation (RAGAS, LLM judges, retrieval and answer quality validation)
• Java (nice‑to‑have)
• Kubernetes (nice‑to‑have)