Data Scientist

At UST, we help the worlds best organizations grow and succeed through transformation. Bringing together the right talent, tools, and ideas, we work with our client to co-create lasting change. Together, with over 30,000 employees in 30+ countries, we build for boundless impact—touching billions of lives in the process. Visit us at UST.com.

Lead AI Engineer
Snapshot
Experience: Overall 8+ years , 5+ years in ML/AI engineering

Education: B.E./B.Tech or M.Tech in CS, Data Science, or related field

About the Role: Build and ship production-grade GenAI and agentic AI applications that automate enterprise workflows — from design through deployment. A hands-on individual-contributor role for a strong builder.

Key Responsibilities
• Build agentic applications using LangGraph, AutoGen, CrewAI, or Semantic Kernel.
• Design RAG pipelines — chunking, hybrid search, re-ranking, memory, and tool orchestration.
• Deploy and monitor AI workloads on Azure (AKS/ARO) with CI/CD and observability.
• Implement responsible-AI guardrails — prompt-injection defense and content filtering.
• Define evaluation metrics for task success, hallucination, latency, and cost.

Must-Have Skills
• 5+ years ML/AI engineering with production LLM/agentic delivery.
• Advanced Python and at least one agent framework (LangGraph, AutoGen, CrewAI, PydanticAI).
• Strong LLM and prompt-engineering skills (GPT, Claude, LLaMA); hands-on RAG workflows.
• Azure AI stack (Azure OpenAI, AI Search, AI Services) and Databricks ML (MLflow, Delta Lake).
• Vector databases (FAISS, Pinecone, Chroma) and embedding/retrieval design.
• Containerized deployment (Docker/Kubernetes, AKS/ARO) and REST APIs / WebSockets / event-driven services.
• CI/CD and version control (Jenkins / GitHub Actions, Git) with SDLC and agile practices.
• Portfolio of 3+ production AI deployments with measurable business impact.

Nice to Have
• Model fine-tuning (LoRA/PEFT), multi-modal AI, and model evaluation frameworks.
• LLMOps / MLOps and model monitoring (drift, latency, cost, hallucination).
• Knowledge graphs (Neo4j); AWS Bedrock / GCP Vertex AI exposure.
• AI-augmented dev tools (GitHub Copilot, Claude Code, Windsurf) for rapid prototyping.
• Enterprise AI security, compliance, and governance awareness.
• Manufacturing or supply-chain domain experience.