brief background

Education: Robotics + Computer Science at University of Pennsylvania

Experience: Recommendation systems at Shoreline/Nvidia, Data Skipping at Databricks, and rust that goes brrr at Mooncake Labs

Research: Co-author Demo at SIGMOD 2025 on ScaleLLM, System Paper at NSDI 2025 on market-based microservice load handling, continued work in UPenn's systems + ML lab


industry

Mooncake Labs - Founding Engineer

#3 hire - stay tuned! data durability, recovery, connectors, talking to customers and more.

Databricks - Intern

Increased performance of one of our largest customers' pipelines by > 500% by helping to skip over large amount of data.

Also started the first young engineer tech talk series, with > 100 attendees for all the talks.

Shoreline -> Nvidia - Intern

Built a recommendation system for runbooks, achieved 95% search recall, explored synthetic benchmarks while they were still nascent.


research

ScaleLLM - SIGMOD 2025

Co-authored a demo at SIGMOD 2025 on ScaleLLM, a system that allows for scaling LLM usage to massive workloads on constrained tasks. Continiuing work on pushing the performance of tiny LMs (<1m params) to generate context-free grammars.

scaleLLM

Rajomon - NSDI 2025

Worked on benchmarking microservice overload frameworks - gnarly Go code for distributed systems.

rajomon


projects

pBFT - asynchronous distributed fault tolerance

Main contributor on a project bringing an asynchronous fault tolerance algorithm from a cyber-physical setting to a data center setting. Advised by Prof. Linh Phan.

github

Sketchy

WIP - A new interface for fine-grained image editing control through sketch + text diffusion models, with only a lightweight guidance adapter required. Continiuing work with UPenn's Computer Vision lab.

sketchy demo | original paper draft