AI Engineer • Data Scientist • Backend Developer
Building AI-powered automation workflows and LLM solutions
Production automation systems and AI-driven operations
LLM training data curation and model evaluation
Full-stack platform with real-time interactions, gamified rankings, and authentication. Built scalable and secure backend (Firebase + REST + WebSockets) using Google Firebase.
Voice-first AI assistant: real-time conversation via WebSocket + Web Speech API + Gemini. Secure auth via Telegram bot; memory management with PostgreSQL for personalized interactions. Impact: Accessible UI, bilingual (ES/EN), privacy-first design for sensitive data.
RAG pipeline: semantic search (FAISS + LaBSE) + PoliGPT for grounded, cited regulatory answers. Automated PDF ingestion: chunking, parallel embeddings, SQLite metadata tracking. Team project (5 contributors); modular design for institutional reuse.
Predictive modeling for urban farming: crop recommendations, cost analysis, risk assessment. Applied AutoML (FLAML) for model selection; built monitoring alerts for production use. Team: 3 developers.
Full-stack typing web integrating LLMs for document analysis. Built modular architecture: TypeScript frontend for responsive UI + Python/FastAPI backend for AI processing. Designed for extensibility: plugin-ready structure for adding new AI features and third-party integrations.
Pipeline combining OCR extraction with LLM-based semantic structuring for kids essay correction automation. Designed for schema compliance and integration with Google Drive and Sheets.
Dashboard visualizing model decisions (SHAP, partial dependence) for stakeholder transparency. Applied statistical analysis to compare model architectures and fusion techniques.
