Proposed an end-to-end hybrid ML pipeline that forecasts football lineups by fusing structured tabular features with LLM-extracted contextual dimensions (injury risk, manager sentiment, and tactical rotation). Architected a monotone-constrained gradient boosting model integrated with a Markov chain for tactical formation forecasting and the Hungarian algorithm for positionally valid player assignments. Evaluated on English Premier League data, achieving 83.9% constrained lineup accuracy and demonstrating a +7.15% accuracy improvement utilizing LLM features over tabular baselines in chronological ablation studies.

Designed an intelligent, context-aware conversational agent using LangChain ReAct agents, dynamically orchestrating custom tools to manage conversation flows without hardcoded logic. Built a state-of-the-art developer assistant chatbot utilizing LangGraph and RAG within a state-machine architecture to autonomously generate code, explain snippets, and detect languages. Architected and deployed a multimodal Toxicity Classification framework for text and images, integrating rule-enforcing guardrails and fine-grained classifiers for robust content moderation.

Architected a RESTful API using Express.js with a modular Routes & Controllers structure to manage communities, posts, and user authentication. Designed and optimized a scalable MongoDB Atlas database using Mongoose schemas for complex relational data (users, threads, comments). Deployed fully containerized microservices architecture, hosting the front end on Vercel and the backend on Render.

Developed a full-stack web application for smart meter management using Java Spring Boot and Angular, facilitating real-time data processing. Implemented serial communication to read meter configurations and write parameter updates directly via COM port connections. Integrated Swagger for robust API documentation and enabled remote device management features (connect/disconnect).

Completed an intensive training program on Cross-Platform Development using Flutter and Dart, focusing on building scalable mobile applications.