At TAMM – Abu Dhabi Digital Government, I built a multi-agent LLM pipeline with structured output validation, fallbacks, and retry logic for production use across 15+ government services.
I designed stateful workflows using LangGraph and Camunda, with human-in-the-loop approval gates. I also implemented LLM observability with Langfuse and a pytest CI suite covering agents and workflows.
Previously, at Amazon, I led the architecture and implementation of a service externalization project that delivered a production RPC API and reduced inter-team dependency latency by about 40%. My experience also includes payments and API development at Amazon, PwC, and eTip, as well as evaluating model outputs for RLHF at Scale AI.

