Agentic AI Engineer
First seen by Alion on Nov 13, 2026.
About the role
As an Agentic AI Engineer at Hudson Data, you will design, build, and ship LLM-powered agents that do real work: researching cases, analysing data, calling tools and APIs, and automating multi-step workflows for our products and for lenders and fintech partners. You will own agents end to end, from prompt and tool design through orchestration, evaluation, deployment, and monitoring in production. You will work closely with our data science, data engineering, and product teams, and help set how Hudson Data builds with AI.
This is a full-time, onsite role in Gurugram, India.
Responsibilities
• Design and build LLM-based agents and multi-agent workflows that plan, use tools, call APIs and databases, and complete multi-step tasks reliably.
• Build retrieval-augmented generation (RAG) pipelines over structured and unstructured data, including chunking, embeddings, vector search, and re-ranking.
• Design clear tool interfaces (function calling, MCP servers) that connect agents to internal systems, data warehouses, and third-party services.
• Build evaluation harnesses: test sets, automated and LLM-as-judge evals, regression tests, and red-teaming, so agent quality is measured, not guessed.
• Take agents to production with solid engineering: APIs, background jobs, retries, cost and latency controls, logging, tracing, and monitoring.
• Put guardrails in place: permissions, human-in-the-loop review, prompt-injection defences, PII handling, and audit trails suitable for financial-services clients.
• Choose the right model for each job across providers (Anthropic, OpenAI, Google, open-weight models), balancing quality, cost, and latency.
• Work with data scientists to combine agents with predictive models, for example agents that investigate fraud alerts or explain credit decisions.
• Keep up with a fast-moving field and bring the useful ideas into our stack.
Required qualifications
• 3+ years of software engineering experience, including at least 1 year building applications on top of LLMs.
• Strong Python, with experience building production backend services and APIs (FastAPI or similar).
• Hands-on experience building agents or LLM workflows: function/tool calling, structured outputs, prompt design, and at least one orchestration approach (e.g. LangGraph, LlamaIndex, OpenAI Agents SDK, Claude Agent SDK, or your own).
• Experience with RAG and vector databases (e.g. pgvector, Pinecone, Weaviate, Qdrant).
• Experience evaluating LLM systems and improving them based on measured results.
• Solid SQL and comfort working with real, messy business data.
• Experience with cloud platforms (GCP or AWS), Docker, Git, and CI/CD.
• Ability to own problems end to end and explain technical trade-offs clearly.
Preferred qualifications
• Bachelor's or Master's degree in computer science or a related field.
• Experience building MCP servers or integrating agents with enterprise systems.
• Experience with observability tools for LLM apps (e.g. LangSmith, Langfuse, Arize, OpenTelemetry).
• Background in fintech, lending, credit risk, fraud, or another regulated industry.
• Experience fine-tuning or serving open-weight models.
• Open-source contributions or public projects with agents or LLMs.
How to apply
Apply on our careers site: https://career.hudsondata.com/careers/agentic-ai-engineer-1004