AI/ML Intern
Six-month internship building two applied ML systems for port operations end-to-end, from dataset collection through deployment handoff.
Truck Container Damage Detection System
Jan – May 2026A full-stack inspection tool (React + Django) combining YOLO-based damage detection with SAM/GroundingDINO segmentation to isolate container regions, plus OCR to read ISO container numbers off the container body. Owned the pipeline end-to-end: dataset collection and annotation with class-balance and augmentation planning, model inference with caching to keep response times down, manual and automatic inspection workflows, and PDF/EIR report generation with damage labels, confidence scores, and bounding-box data. Also handled backend input validation, error handling, and a multilingual, theme-aware UI. Completed and handed over to the supervisor in May 2026.
SLPA Chatbot
May – Jul 2026A retrieval-augmented generation (RAG) chatbot serving separate internal (staff) and external (public) flows in English, Sinhala, and Tamil. Built the full pipeline solo: query processing, embedding-based retrieval over an indexed knowledge base, and prompt-engineered response generation tuned for concise, official-toned answers, with small-talk handling as a fallback for greetings and general messages. Also shipped an attendance-request feature (form + backend), a monitoring panel for chat and attendance activity, and structured logging for debugging failed responses. Delivered on React + Vite (frontend) and FastAPI (backend), from requirements gathering through final testing and demo.