At Data Spark, I build AI-powered tutoring and assessment workflows for TNPSC, Sasthra, JEE, NEET, and school learning. I've developed RAG-based WhatsApp tutoring, multilingual question generation, AI validation, OMR automation, and reusable AI infrastructure.
I reduced CBT question-generation runtime by 70% and API costs by 50% through workflow consolidation, execution analysis, fallback mechanisms, monitoring, and stress testing. I also work with persistent student sessions, ATS resume scoring, visa-interview workflows, speech-to-text, MongoDB, vector databases, and Gemini/OpenAI-backed systems.
Previously, at BLP Industry.AI, I developed computer-vision safety-detection models using YOLOv8, PyTorch, OpenCV, Roboflow, and CVAT. I processed and annotated 10,000+ images and improved mAP@50 by 30% across 15,000 CCTV frames while maintaining real-time performance.
