Role & responsibilities:

- Hands-on builder first. This is not an architecture-only role the role is expected to write production code alongside the team, lead by doing, and be the most technically capable person on the Agent Skills Team.
- LLM Foundations deep understanding of transformer behaviour, tokenization, context windows, sampling controls, hallucination patterns, and latency/cost tradeoffs; uses this knowledge to make sound model selection and prompt architecture decisions
- Production-grade prompt engineering task decomposition, system/developer/user prompt layering, reusable templates, guardrails, adversarial robustness testing (jailbreak resistance, edge-case handling), and prompt versioning with experiment tracking and rollback
- AI evaluation & quality engineering designs eval frameworks with metrics covering accuracy, groundedness, safety, latency, and cost; builds automated eval pipelines (LLM-as-judge, golden sets, regression suites); owns observability including trace logs, failure clustering, and drift detection
- Agentic AI & A2A orchestration — hands-on implementation of agent orchestration, multi-step reasoning, ReAct / plan-and-execute; designs A2A patterns (agent roles, delegation, handoff protocols, state boundaries) and multi-agent governance (permissions, execution constraints, escalation paths)
- MCP (Model Context Protocol)



— builds and owns common MCP servers; defines tool schemas, capability discovery, and interface contracts; implements tool reliability patterns (input/output validation, fallback strategies, circuit breakers)
- RAG pipeline implementation — builds chunking, embedding, and retrieval pipelines end-to-end; tunes retrieval quality through code; owns knowledge base ingestion and refresh strategy
- Software engineering & MLOps backbone — Python/TypeScript proficiency; CI/CD for AI systems with eval gates before deployment; cost/performance optimisation (caching, batching, model routing); data security (PII handling, secrets management, audit logging)
- Cross-team technical leadership — Leads by example in code quality and AI engineering practices; can explain model behaviour and limitations to non-technical stakeholders

Key Skills:

- Python (production-grade)
- LLM application development
- Agentic AI & multi-agent orchestration
- RAG implementation
- MCP (Model Context Protocol) or equivalent tool-use protocols
- Production-scale Prompt Engineering
- AI evaluation, testing, and observability
- Robust software engineering and MLOps practices

Nice to Have:
TypeScript, Neo4j, SAP AI Core, Financial Services domain experience

📌 Senior AI Engineer (Chennai)
🏢 KG Information Systems (KGISL)
📍 Chennai