At GEP Worldwide, I built a RAG pipeline using SLMs, Docling, and Tesseract for semantic clause extraction in Legal Assistance. I migrated it from Azure ML to in-house infrastructure, cutting cloud costs and reducing document-processing latency by 87%.
I've also built a Node.js and MySQL backend for Education New Zealand's high-volume session tracking, with RESTful APIs, low-latency reporting, modular services, and Postman-driven testing. At VVDN Technologies, I developed an encrypted Python terminal chat system with RSA/AES, RBAC, and authenticated UDP service discovery.
I enjoy building from fundamentals, including a C-based language compiler and a digit classifier that achieved 99.7% accuracy. I also placed second in the ContraShieldAI intern hackathon for a multi-agent contract-analysis system.
