Machine Learning & Data Team Lead — Arivihan Technologies
Fine-tuningLed LLM and RAG-focused model training and evaluation for AI products, including fine-tuning domain-specific LLMs for educational use cases. Built an AI model evaluator to benchmark models on accuracy, latency, and domain relevance to support iterative training and selection. Developed and deployed AI chatbots using Rasa with Flask/FastAPI and OpenAI/Gemini integrations to improve response behavior through retrieval and tuning. • Fine-tuned domain-specific LLMs using Hugging Face and OpenAI fine-tuning APIs • Built an LLM evaluation framework (accuracy, hallucination rate, response quality) • Orchestrated RAG pipelines and LangChain-based workflows for training/evaluation cycles • Implemented chatbot systems that combine fine-tuning and RAG for improved answers