At Lifebox Next Gen, I developed GNAN AI, a local SLM-based adaptive interviewer with resume, role, and HR interview modes. I built its evaluator and question generator logic for adaptive follow-ups, difficulty adjustment, and structured candidate assessment.
I also benchmarked SLMs and optimized KV cache, prompt cache, and context handling for latency and memory efficiency. In my projects, I built a RAG document assistant that processed 500+ PDF pages and a real-time American Sign Language recognition system.
