At Y77, I develop Python automation systems that integrate APIs, databases, and third-party tools. I also optimize pipelines with error handling and logging to improve reliability and performance.
At Chessworld.ai, I created a dataset from tournament chess videos and trained a YOLO model to detect chess pieces and board states. I designed a hybrid inference pipeline that combined YOLO predictions with chess-rule logic to convert video into valid PGN move sequences.
At UNISYS - Campus Connect, I converted videos to transcripts using Whisper and built a RAG pipeline with LLM fine-tuning to extract trends and business insights from 10K+ records, reducing manual analysis by 30%. I also delivered Python and FAISS vector search systems for transcript retrieval, LLM-powered summarization, reporting, and stakeholder dashboards.
In my projects, I implemented a multimodal anti-drone surveillance pipeline and created LLMTestBench to compare RAG and agent outputs across prompt, model, and retrieval changes. My research review on machine learning and deep learning models for brain tumor analysis using MRI was published by Springer.
