Josh T

I’m a Software Engineer delivering end-to-end AI/ML systems in production—especially RAG, LLM evaluation, and secure deployment. At Bytronic Vision Intelligence, I translated client requirements into actionable technical specifications and secured over £1M in project funding.

I architected and deployed a production RAG system using LangChain with hybrid retrieval and optimized chunking, then built an evaluation framework with deepeval (recall, precision, faithfulness, and contextual relevancy). I improved average answer correctness by 30% and answer relevancy by 5%, while also deploying a FastAPI gateway with rate limiting and API-key authentication and streaming support via vLLM.

I’ve also shipped practical AI solutions at scale: an LLM-powered expense filing assistant that reduced administrative time by 60% and processed 1000+ transactions monthly, plus a 24/7 RAG chatbot that cut support tickets by 40%. My background extends to model deployment with Docker, vector DBs (Qdrant, ChromaDB), CI/CD with GitHub Actions, and computer vision work (YOLO models achieving 1mm positional accuracy); I reinforce this with project work like my LLM Integrated Ebook Manager and a dissertation focused on fault classification from gearbox acceleration data.