At Moveinsync, I designed and deployed an AI-powered conversational chatbot integrated with Jira APIs. It achieved 80% intent classification accuracy and autonomously resolved 60% of support tickets.
I enhanced the assistant with LangGraph, LangChain, and BERT, applying reinforcement learning with human feedback to improve classification and resolution. I later migrated the solution to N8N for workflow automation across teams.
In the NVIDIA Nemotron Model Reasoning Challenge, I fine-tuned a 30B-parameter reasoning model using LoRA and synthetic datasets, achieving a reported evaluation score of 0.84. I’ve also built machine-learning projects for text classification and tele-consulting.
