MANSI GOEL

Contributed to a large-scale multimodal video search agent, owning ingestion pipeline, search agent parts, eval framework, and Claude skills integration. Designed multi-agent orchestration pipelines using LangGraph, LangChain, and NeMo Agent Toolkit, and developed computer vision workflows with SAM2 and depth estimation.

Designed and deployed end-to-end ML pipelines for ultrasound image analysis, achieving clinically validated accuracy benchmarks. Authored a research paper accepted at ICHI 2020, selected as Best Paper in the Analytics Track.

Led end-to-end development of AI/ML products including a Sentiment Analysis engine and a Software Recommendation System deployed in production. Designed and implemented a Reinforcement Learning-based recommender system and built scalable ETL data pipelines.

Contributed to Camera application development, enhancement, and legacy code maintenance for Android smartphones. Diagnosed and resolved cross-model camera bugs, improving device quality scores and reducing post-release defect rates.

Activities and societies: Built and published Backtracker on Google Play Store as part of a startup project during the program.

Pursued a Master of Science in Computer Science, during which a startup project led to the development and publication of Backtracker on Google Play Store.

Activities and societies: Published a paper in IJCA: 'Rank Based Retrieval of Answers for Noisy Queries.'

Completed a Bachelor of Science in Computer Science, and published a paper in IJCA titled 'Rank Based Retrieval of Answers for Noisy Queries.'

Completed the DeepLearning.AI TensorFlow Developer Specialization on Coursera.

Completed the Machine Learning course by Andrew Ng on Coursera.