As a backend engineer on the platform team:* Hunted down a performance bottleneck to unblock app scaling. Decreased response time on commonly used endpoints from around 600ms to 200ms, leading to a noticeable increase in front-end responsiveness. Reduced deployed API nodes by 65%, cutting our team’s cloud bill by around 50%, and reducing the overall cloud bill by around 30%.* Improved database usage while maintaining high-traffic but cumbersome legacy code. Identified sequential requests that could be batched, or repeated requests that could be cached. ex: reduced 40+ database queries in a single request to about 5.* As part of a suite of broader platform improvements, developed a new Python API using asyncio and gRPC, also replacing questionable use of NoSQL with a Postgres database.As lead backend engineer on the "wallet" team:* Delivered, from initial design to release, backend endpoints and a series of PubSub workers that integrated with a third-party banking provider, for the new “wallet” feature in our application.* Maintained end-to-end tests for each story delivered, allowing the team to focus on fixing edge cases and finishing front-end work during the final months of development, with zero backend regressions.* Pushed to move up integration with the front end before UI work was complete, exposing issues in our workflow that required new designs, avoiding weeks of potential engineering re-work.* During launch, wrote a support document and ran fire drills on common technical issues. Organized direct communication with customer support to flag issues, fixing all internal issues same-day. Post launch, released tools that support could use to fix issues without engineering involvement.* Created OpenAPI documentation that replaced hand-written wiki docs (which were often under-specified and out-of-date), with CI deploying the docs to a web page via Google Cloud. This streamlined communication and reduced re-work for the front-end team.

Migrated 1 year of Twitter data into Elastic Search using Spark, allowing the search team to tear down an expensive Lucene index while drastically simplifying ingest API code.Translated AI python projects into ingest services, merging and loading multiple data sources into a graph database, then learned and demoed Neo4j, enabling other teams to begin prototyping new features.Designed and begun adoption of ingest strategies that would allow for one-click rollback/re-ingesting of data, mitigating time-consuming procedures when data was lost or needed correction.Worked directly with DevOps to issue pull requests to their code, unblocking my team when DevOps had a heavy workload.Performed a deep dive into our codebase, teaching team about project internals while allowing us to remove endpoints other teams no longer used, greatly simplifying maintenance.

On the data team, I learned Scala and Spark from scratch in order to lead the charge on adopting these tools to augment our legacy MapReduce batch processing. I added tooling to support reading Avro models in our nightly MapReduce and Spark jobs, allowing us to break up simultaneous, company-wide code updates that would often result in failed nightly processing.On the API team, I acted as the resident expert on Hadoop, MapReduce, and Spark, while I learned how to work with microservices containerized in Docker and deployed onto KubernetesCreated an internal style and standards guide for the platform’s Avro data models. Worked with teams that had competing design goals, and promoted an understanding of guideline tradeoffs.

As an Operating Systems TA, I lead 2 one hour sessions introducing the C programming language. I designed my own lectures that guided students familiar with Java or similar languages through the parts of C that might provide the most trouble, while introducing other techniques they would need to know to complete their assignments. I taught other recitations sections, and graded tests and assignments.I also served as a peer tutor for introductory computer science and computer engineering courses. I emphasized understanding how basic concepts pieced together, and helped students lead themselves through problems.

Software Engineering Intern in the Energy Efficiency section of the Kernel Platform Group, working on a Connected Standby feature.Work involved C development which worked largely within a Windows service. I used WinDbg and Xperf on virtual and physical machines to demonstrate the correctness of my design. I used a power meter attached to a physical machine to demonstrate the situations in which my project could result in lower power consumption, and proposed possible ways in which my project could be extended.

Software Development Engineering Intern on the Build Automation and Test team working on an internal automation and translation tool.Worked collaboratively with a team of six interns that needed to interact with and learn from employees across the company. Applied separation of policy and mechanism by writing a small framework for the project that enabled others on the team to more rapidly increase the number of targets our translation tool worked with.