Jan 2024 – Jan 2026 · 2 yrs · San Francisco, California, United States · Hybrid

• Worked on Daft, the data engine for AI and multimodal data, written in Rust+Python
• Joined as employee #4 during seed-stage, raised $30M Series A and powered petabytes of AI data processing daily
• Led projects on core engine capabilities, including: LLM batch inference infra & optimization, distributed joins, cost-based query optimizer, SQL execution, and integrations with Iceberg, Delta Lake, Hugging Face
• Speaker at Ray Summit, SF Iceberg Meetup, MDS Fest

Jun 2023 – Aug 2023 · 2 mos · Boston, Massachusetts, United States · On-site

• Analyzed large, textual datasets and developed signals using NLP/language models and statistical techniques
• Built and utilized code profiling tools to significantly improve runtime and memory usage of datastore querying functions used commonly across the firm
• Worked on the Quant Research team, used technologies including Pandas, Dask, Huggingface Transformers, AWS, SQL

Aug 2022 – May 2024 · 1 yr 9 mos · Berkeley, California, United States

• 2023-24 - CS 189 (Intro to Machine Learning)
• 2022-23 - CS 70 (Discrete Math and Probability)
• Taught discussion, ran office hours, wrote content, and handled logistics for EECS courses at UC Berkeley
• Ran some of the most popular courses at Berkeley, serving hundreds of students per semester

Jun 2022 – Aug 2022 · 2 mos · Chicago, Illinois, United States

• Developed tooling to effectively process and analyze massive amounts of real-time data from cryptocurrency exchanges
• Worked closely with traders on the cryptocurrency desk to interpret market data and parallelize data processing
• Used technologies such as Python, C++, Databricks/Apache Spark

Jan 2022 – Dec 2023 · 1 yr 11 mos · Berkeley, California, United States

• Led the 20 member team behind berkeleytime.com, the most widely used course discovery platform at UC Berkeley, with 20k+ monthly active users
• Coordinated efforts between frontend, backend, and design teams to deliver new features to students
• Worked with technologies such as Django, Kubernetes, Node.js, React, PostgreSQL, MongoDB, GraphQL, Figma

Sep 2020 – Dec 2023 · 3 yrs 3 mos · Berkeley, California, United States

May 2021 – Aug 2021 · 3 mos · Sunnyvale, California, United States

• Built tools to analyze database usage and availability for LinkedIn’s internal document data store service
• Utilized data from time series graphs, hardware usage capacity, and numerous sources to determine database health
• Developed a full-stack application to automate database analysis using Python Flask and Ember.js

Feb 2021 – May 2022 · 1 yr 3 mos · Berkeley, California, United States

• Worked on projects in the search for extraterrestrial life in the universe
• Built tools for researchers to scalably run machine learning algorithms on massive radio telescope datasets
• Deployed hybrid cloud Kubernetes and Slurm clusters optimized for SETI research workloads

Jun 2020 – Aug 2020 · 2 mos · San Francisco, California, United States

• Led a project to transition the Cloudflare Dashboard from a PHP generated site to entirely statically served React
• Converted the Cloudflare API schema to OpenAPI using Typescript as part of work to rehaul the API documentation
• Built features used directly by all Cloudflare users using technologies such as Nginx, GCP, Cloudflare Workers

Jun 2019 – Aug 2019 · 2 mos · San Francisco Bay Area

• Developed and refactored user interfaces for the Docker Trusted Registry, a private image registry for enterprise
• Used technologies including Javascript, React, and Cypress to write high-quality frontend code
• Work deployed to and directly used by hundreds of companies around the world

Jul 2018 – Aug 2018 · 1 mo · San Francisco, California, United States · On-site

• Built tools to ultimately help those in developing countries grow their businesses and improve their lives
• Created a browser extension to improve the user experience of Kiva translators using HTML, Javascript, jQuery
• Learned to effectively communicate technical details to nontechnical users, worked with a high feedback loop