As a Machine Learning Engineer at Medeloop, I am currently leveraging my expertise to develop and optimize machine learning infrastructure that accelerates early-stage clinical research.
Previously, I was a Startup Solutions Architect at Amazon Web Services, working with Generative AI companies across NAMER to build out large GPU clusters, advise on AI/ML best practices, and generate prototypes for key strategic customers. I also got the opportunity to work with an amazing AWS Ops team to ship an internal tool called salesparty to AWS sellers globally, which reduced seller's research time from hours to seconds.
Some of my favorite papers include [1][2] and [3]. I am especially interested in the effects of reinforcement learning methods on polysemanticity and superposition in large neural nets.