Agentic AI Engineer — Data & Analytics

Agentic AI Engineer — Data & Analytics

5-12 yrs of exp

About the Role

We are building a portfolio of AI agents and reusable accelerators that transform how enterprises run their data, analytics, and supply chain operations. As an Agentic AI Engineer, you will design and build production-grade AI agents — systems that reason over enterprise data, invoke tools and APIs, and execute multi-step analytical workflows with appropriate guardrails. This is a software engineering role first and a GenAI role second: you will ship code that runs in client production environments, not notebooks that run-in demos.

What You Will Do

• Design and build LLM-powered agents for data and analytics use cases: data quality triage, pipeline diagnostics, analytics copilots, document intelligence, and domain-specific decision agents etc.
• Implement agent orchestration using frameworks such as LangGraph, CrewAI, Claude Agent SDK, Semantic Kernel, or equivalent — and know when to use none of them.
• Build robust tool/function-calling layers over enterprise systems: SQL engines, REST APIs, data catalogs, ERP/planning systems, and vector stores.
• Design and implement retrieval architectures (RAG, hybrid search, GraphRAG, semantic caching) with measured retrieval quality, not assumed quality.
• Build evaluation harnesses: golden datasets, LLM-as-judge pipelines, regression suites for prompts and agent behavior; treat evals as Continuous Integration, not as an afterthought.
• Implement guardrails and safety controls: input/output validation, PII handling, cost and latency budgets, human-in-the-loop checkpoints for consequential actions.
• Productionize agents: containerization, CI/CD, observability (tracing every agent step), versioning of prompts and models, rollback strategies.
• Collaborate with developers and solution architects to ensure agents consume governed, well-modeled data — not raw chaos.
• Presents demos; handles technical Q&A

Must-Have Qualifications

• Strong software engineering foundation in Python (typing, testing, packaging, async); working proficiency in SQL.
• Hands-on experience building LLM applications beyond prototypes: at least one system with real users, real failure modes, and real iteration.
• Practical understanding of LLM behavior: context management, structured outputs, tool calling, prompt versioning, token/cost economics.
• Experience with at least one vector database or hybrid retrieval stack (e.g., pgvector, Pinecone, Weaviate, OpenSearch, Azure AI Search) and the trade-offs between them.
• Experience deploying services to at least one major cloud (AWS, Azure, or GCP); comfort with Docker and CI/CD pipelines.
• Ability to reason about non-determinism: designing systems that fail gracefully and are testable despite probabilistic components.
• Sound judgment on when a problem warrants an agentic solution versus a deterministic one, with clear rationale for the choice.Good to Have
• Experience with agent evaluation frameworks (Ragas, DeepEval, Braintrust, LangSmith/Langfuse or equivalent).
• Exposure to MCP (Model Context Protocol) or building tool integrations for AI systems.
• Knowledge graph or semantic layer experience.
• Domain exposure to supply chain, manufacturing, or enterprise operations data.
• Contributions to open source, technical writing, or internal accelerator/IP development.