At LexisNexis RELX, I architected and deployed an internal AI Evaluation & Performance Platform on AWS ECS, with a Python backend and web observability dashboard. It included automated test scheduling, LLM-generated synthetic test cases, and real-time evaluation metrics.
I automated evaluation of an end-to-end RAG pipeline across ingestion, retrieval, and synthesis. Using hybrid search and cross-encoder re-ranking, I improved Context Precision from 68% to 89%.
At Zycus, I designed distributed-system test suites for event-driven microservices and built a Python and LLM assertion service in CI/CD pipelines. That work cut manual triage by 90%; my high-concurrency load testing also helped cut p90 latency by about 30%.
I also engineered Playwright, Selenium, and Cucumber UI and API suites that reduced regression bugs by 40%. In personal projects, I developed an LLM-powered bug reporting system and a ticket triage service that generated preliminary root cause analyses and routed 20–30 actionable tickets per day.

