Darren Udaiyan — Software Developer & Physicist

About

I have extensive experience in QA and Development across many commercial environments: oil, CAD, legal accounting, defence, electricity and software management. I have created automation frameworks for testing web, desktop and APIs, as well as bespoke tools to aid QA, developers and the support team. I hold a patent on QA processes and have published over 10 papers as part of my academic career.

Patent · US8024705 Published papers

Open Source

Two public projects on my GitHub profile.

intelligent-test-platform

Python 3.12 · C++17

An AI-assisted quality-engineering system that turns raw test results into actionable insight: LLM-assisted failure triage and root-cause clustering, multi-signal flakiness scoring, regression-prevention gating (it fails CI if a fixed bug reoccurs), test-effectiveness metrics, and autonomous triage agents producing prioritised action plans. A modular pipeline (ingestion → analysis → flakiness → regression → metrics) runs over JUnit/pytest XML. It is offline-first, deterministic in CI, fully type-safe (Pydantic, mypy strict), with ~93% coverage, plus a C++17 accelerator via pybind11 and a pure-Python fallback.

porfolio-demo — AI Core Services QE Portfolio Demo

C# / .NET 8 · some Python

A production-shaped demonstration of quality engineering for AI/LLM services: testing non-deterministic AI deterministically using “seams” that swap real models for deterministic fakes, removing flakiness, secrets and cost from testing. It is a minimal ASP.NET Core service (/chat, /rag-query) with 122 tests across functional, contract, AI-evaluation, fairness, security and performance, offline evaluation metrics and safety screens (PII, toxicity, prompt injection), plus threat modelling and compliance mappings (NIST AI RMF, EU AI Act).