yash jangid

Designed and implemented production-ready constraint-based scheduling solvers and hybrid decision pipelines combining symbolic reasoning with learned models to improve robustness and correctness. Built automated validation and data-curation frameworks and integrated solver pipelines into production workflows.

Led end-to-end development and production deployment of a multimodal voice-analysis pipeline (ASR, diarization, emotion detection, summarization), achieving 95% transcription accuracy and real-time multilingual processing.

Improved predictive model accuracy by 20% through model selection, HPO, and feature engineering; automated image-processing pipelines for detection and segmentation and deployed scalable ML inference APIs with CI/CD. Designed optimized SQL for ETL and supported retraining workflows.

Led development of a personalized grocery recommendation system using collaborative, embedding DNN and sequence models on 100K+ transactions; deployed FastAPI real-time serving with Redis and CV-based visual search, achieving 78% test accuracy and +35% engagement (10K+ DAU).

Activities and societies: Led ML-focused projects including SmartShop recommendation system and a multimodal voice analysis pipeline; contributed to related publications.

Completed a Bachelor of Science in Information Technology with coursework and projects focused on machine learning, computer vision, and full-stack ML deployment.