New York, New York, United States
I'm a graduate of the University of Pennsylvania with dual degrees from the Wharton School of Business and the School of Engineering and Applied Science.
- Led the rearchitecture and rebuilding of the entire product backend to 1000x the portfolio construction performance, implement accurate backtesting, and create an experimentation/exploration environment for the quant team - Upleveled the operational maturity through a focus on reliability, fuzzing/randomized testing, and regression testing on production data. - Partnered with recruiting to 3x the engineering team with heavy contributions from personal network referrals - Spun up the ML team
I took Clearstep from just an idea to a product used by tens of thousands of users across multiple geographies, hospital systems, and insurance companies. By the end of my tenure, the company was bringing in over $1M ARR and had a significant sales funnel built up. Along the way, I: - fell in love with operating as a small, high-performing team - personally built an infrastructure and product backend that passed security and reliability review to serve traffic for CVS Health, United Healthcare, HCA Healthcare, etc. - hired and ran an engineering team of ~5 - developed pricing strategy and helped negotiate 6-7 figure deals with the largest healthcare companies in the U.S.
- Led a team of ~5-10 engineers of varying skill-levels. Worked to develop mentorship and technical leadership opportunities for more senior members of the team, while also ensuring junior members received mentorship and support. The team successfully implemented several projects such as a profiling system and a new time-series monitoring system, while maintaining existing systems. During this time, the team doubled in size while increasing diversity. - Led the design and implementation of a server monitoring system (https://dropbox.tech/infrastructure/monitoring-server-applications-with-vortex). The new system had 6x better data resolution, up to 1000x faster queries, drastically reduced availability issues and data loss, cost 2x less to operate, and removed all manual maintenance operations that the previous system suffered from. Coordinated and led the migration effort across hundreds of services, culminating in the deletion of the previous system and client libraries. As part of this project, made numerous improvements to the internal RPC framework and the open-source grpc-go project, reducing memory allocations and leading to throughput and max-QPS wins of up to 10% for all infrastructure services. - Introduced dead code detection for Python and CSS. Deleted millions of lines of dead code and tests, simplifying the product, improving performance, and reducing Continuous Integration costs. - Introduced a Python test profiler into our build system and fixed a myriad of performance issues across test collection, test environment setup, backend services, reducing the time spent on running a no-op test integration test from 50 to 5 seconds - Introduced lighter-weight test helpers for core flows and switched most tests all across the codebase to using them, resulting in reduction of CI costs by 2x.
- Hand-picked for elite company-wide effort to speed up the website by over 2x. Optimizations ranged from speeding up DB queries across many product features to improving the serving layer to reducing asset sizes to speeding up frontend rendering. Personally responsible for improvements of up to 1s page load times across many pages. At the culmination of the project, we distilled our best practices to performance guides and shared with product teams. - Introduced a Remote Procedure Call framework based on gRPC (https://dropbox.tech/infrastructure/courier-dropbox-migration-to-grpc) to unify existing inter-server communications across all languages. Worked with upstream to diagnose issues/regressions and contributed patches. The framework improved security, availability, ergonomics, and performance over the existing solutions. Led a migration effort to migrate hundreds of services across the organization to the new framework. Wrote and open-sourced a Python type stub generator for protocol buffers (https://github.com/dropbox/mypy-protobuf) to improve type safety for Python applications using RPC. - Sped up document preview generation over 2x and optimized storage to reduce costs by millions of dollars per year. - Simplified and improved our approach to sandboxing image thumbnail generation to improve throughput 4x while reducing maintenance toil. - Instituted static analysis and dead code detection for Golang which yielded some easy fixes for lurking concurrency bugs. Deleted resulting dead code across many infra services.