At Mecgale Pneumatics, I built and deployed IT Ticketing and Visitor Management systems as cost-effective alternatives to ServiceNow. I managed their software development lifecycle, from requirements analysis and database design through full-stack development and deployment.
During my AI/ML Virtual Internship at EduSkills Foundation (Google), I integrated Google Cloud Vision API for OCR-driven document processing and information extraction. I also developed API-based data processing workflows and applied model quantization to reduce model size by 60% while preserving accuracy.
My projects include a FastAPI and PostGIS pipeline for repairing and normalizing farmland GPS datasets, and an AI Meeting Intelligence Assistant for transcription, summarization, and information extraction. I also developed a QLoRA fine-tuning pipeline for MedGemma-4B, evaluating it across 15 chest X-ray diagnostic categories.
