• Currently working on a startup which sells data analysis services using language models (LLaMA & GPT-4) to generate andexplain data pipelines that can be deployed to AWS using Lambda, Sagemaker, and EC2 for compute• Built software using Python, Ray, and LLaMA 2 which logically understands and corrects data with errors• Built software that allows language models to create and iterate on data pipelines which can be dockerized and deployed• Meeting weekly with clients to iterate on requirements and provide updates in an agile-style workflow• Creating the site for Qtack using React and NodeJS.

• Co-founded a startup that made codebase context-aware completion tools for programmers using OpenAI’s API (https://point-drift.com/)• Fine tuned the LLaMA 2 language model on coding datasets using QLoRa, and fine-tuned GPT-J using the ROME algorithm.• Conducted hundreds of interviews with programmers on pain points and ways to improve their workflows.• Oversaw pre-alpha testing of an MVP which used a Django backend connected with a React frontend.• Developed custom code-indexing solutions to allow LLMs to pull more relevant code

• Used TensorFlow to create a neural network that quickly assessed the image quality of large (gigabyte-size) pathology slides• Coordinated standups and updates between members• Created an automated model analysis pipeline using SQL and SQLite (a relational database) to store test results