Yasas User

I’m an AI/ML engineering intern and Computer Science undergraduate with a deep interest in Systems Programming and Generative AI. I’m motivated by building complex architectures—from low-level components to production-grade AI systems—and I enjoy turning ideas into dependable software.

I’ve engineered production-grade Retrieval-Augmented Generation (RAG) systems using Python and modern tooling like LangChain and LangGraph. In my work, I combine Hybrid Search (semantic vector search with ChromaDB plus lexical BM25) and improve answer quality using Cross-Encoder reranking and LLM-driven query expansion with Google Gemini and Ollama. I also built automated evaluation pipelines using the Ragas framework to measure faithfulness and correctness against a golden dataset.

Beyond RAG, I’ve built multi-agent research mapping systems that autonomously fetch, analyze, and connect research papers. Using specialized agents for metadata ingestion (ArXiv, OpenAlex, CrossRef), semantic link weaving, and graph visualization, I delivered an interactive Streamlit dashboard with an LLM-powered chat interface for research discovery.

I balance applied ML with strong systems fundamentals. I engineered a lightweight mini kernel from scratch in C and Assembly (including GDT and IDT initialization) and built core drivers for screen output and keyboard input. I’ve also developed a real-time emotion recognition pipeline with TensorFlow and OpenCV, and I enjoy shipping across platforms, including a React Native/Expo gym leveling app with clean state management via Zustand.