Siddharth Sharma

I am an undergrad studying CS and Math at Stanford. 🌲

I'm currently building search and middle-layer applications that bridge large language models with knowledge bases. I also like exploring frontier ML primitives like meta-learning, reinforcement learning, and differential privacy. In the past, I've worked on semantic search, relational embeddings, computer vision for healthcare, and quantum ML. During high school, I was fortunate to work on researching biological neural networks and clinical object detection with Professor Emeritus Bernard Widrow and Dr. Daniel Rubin respectively. I grew up in the San Francisco Bay Area.

If you want to get in touch, email me at sidshr@stanford.edu.

Projects

I've worked on formal research projects focused on adaptive biological learning (Hebbian-LMS), seizure patient tracking for patients, and compressing and pruning GPT-2. Here's a selection of some of my projects:

• Plato: AI pair-programming platform powered by your voice
• Cardinal Compass: AI-powered semantic search for Stanford courses
Over 50,000 search requests and over 2,700 unique student upvotes at launch
See Stanford Daily article

• LearnEngine: GPT-enabled books & media recommendations
• Polymath: smart summary educational search engine
• GPT-or-Not: a probabilistic method to detect text written by LLMs
• SmoothShot: VAE-based restoration of sports highlights
  
• MarkovMatch: Simulating Tennis Matches with Absorbing Chains
• Cerbo: Abstracting ML production for Beginners (15,000+ downloads!)
• A Guide to ML Applications: lead author of 283 page technical textbook on machine learning.