Adam Khakhar
AI entrepreneur, researcher, and technical leader acquired by Hebbia
New York, New York
Summary
Adam Khakhar is a serial entrepreneur with a track record of identifying market needs and building innovative AI-driven solutions. He co-founded FlashDocs, an API-first platform that transformed AI prompts into enterprise-quality slide decks, leading to its acquisition by AI giant Hebbia within a year. This venture demonstrated his ability to rapidly develop and scale a product that solved a significant pain point for businesses. yahoo+1
With a strong academic background, Adam has pursued doctoral studies in Machine Learning at New York University, complemented by a Visiting Doctoral Scholar affiliation at Stanford University. His research interests lie in Deep Learning and Optimization, evident in his published works on topics such as PAC prediction sets for large language models of code and delta hedging in automated market makers. google+1
Adam possesses significant technical expertise, having served as CTO of FlashDocs and as an Engineer at prominent firms like Meta (Facebook A.I. Research) and The D. E. Shaw Group. His hands-on experience spans from developing AI-powered systems to co-founding a quantitative hedge fund, Manifold, where he led machine learning efforts, showcasing a blend of theoretical knowledge and practical application in complex financial and technological domains. techsavvy+1
Adam's entrepreneurial journey began early, starting a streetwear brand (PureNYC) and an ed-tech app during his high school years. This early exposure to building ventures from the ground up fostered a keen sense of innovation and business acumen, which he later leveraged in co-founding a quantitative hedge fund and ultimately FlashDocs. yahoo
Work
Education
Projects
Writing
PAC Prediction Sets for Large Language Models of Code
January 1, 2023This paper presents a strategy for quantifying uncertainty in deep neural networks for structured prediction problems like code generation, proposing a method to represent prediction sets as partial programs with theoretical guarantees. It was presented at the 40th International Conference on Machine Learning (ICML) in 2023.
Delta Hedging Liquidity Positions on Automated Market Makers
January 1, 2022Authored a paper on delta hedging strategies for liquidity positions in automated market makers, presented at the Crypto Economics and Security Conference in 2022.
Neural Regression for Scale-Varying Targets
January 1, 2022Co-authored a paper discussing neural regression techniques for targets that vary in scale, published as an arXiv preprint in 2022.