Generative AI Engineer
ShimentoX Technologies · Sep 2025 – Oct 2026
- Uber Enterprise AI RL Gym / Agent Evaluation Platform: Developed an enterprise evaluation workflow that converts API and business-domain knowledge into runnable Gym environments for testing and grading AI agents.
- Contributed to the Gym Factory pipeline using RAG, LLM-based multi-source research, API knowledge grounding, capability modeling, synthetic data generation, and REST/MCP tool integration for AI agent evaluation.
- Implemented evaluation flows validating tool/API calls, workflow sequencing, database/application state changes, expected outcomes, final responses, constraints, and failure patterns across agent trajectories.
- Worked on safety and governance controls including approval checkpoints, evidence/provenance handling, gap analysis, deterministic checks, and state-aware grading.
- Built agentic workflows using LangChain, LangGraph, MCP servers, and tool/function calling to coordinate multi-step tasks, integrate REST APIs, and validate agent execution against expected business outcomes.
- Integrated Azure OpenAI, Azure AI Foundry, and Azure AI Search into enterprise AI workflows for LLM inference, retrieval, and knowledge-grounded application development.
- Implemented LLM/RAG evaluation and deployment workflows using automated quality checks, groundedness and relevance assessment, Docker, CI/CD, and model lifecycle tooling to support reliable, traceable enterprise AI applications.
- HSBC - Spectra / Test-Genie: worked on an AI-powered API testing platform that parses OpenAPI specifications and generates BDD/Gherkin test suites, including functional/non-functional scenarios, automated execution, and HTML/JUnit reporting.
