Offloaded event logging data from application pods to S3, streamlining rolling blue/green upgrades by lowering the copied data needed for a new live cluster and decreased storage costs.

Leveraged S3, DuckDB, S3Proxy, and optimized compression methods to transform a live cluster with 200+ million events into a compacted 8.4 GB, achieving a 24x cost reduction per GB compared to managed PostgreSQL storage service while maintaining service provider agnosticism the use of Azure Blob Storage in addition to S3. DuckDB is used for querying from data sources like S3 using normal SQL statements.

On the Platform Team, I worked on separating our monolithic applications into microservices run in a Kubernetes Cluster. This enables the application to scale individual components independently when they are under load.

Implemented Kubernetes observability enabling production cluster insights via Kubernetes Metrics such as Pod and Node CPU + Memory utilization enabling insights issues without having to access the production Kubernetes cluster.

Worked as a full-stack engineering intern with a primary focus on developing a user dashboard aimed at empowering network administrators to visualize network traffic and proactively address network anomalies. This multifaceted role encompassed various responsibilities:

Front-end: Development using React, charting libraries, and breakpoints, ensuring an intuitive and visually appealing user interface for widgets on the dashboard.

Backend: Using Ruby on Rails, I enhanced system's functionality and performance by building out an optimized data pipeline to handle data processing.

Database: Gained extensive experience in PostgreSQL, harnessing its capabilities to efficiently store and manage data.

Big Data Processing: Leveraging advanced algorithms and sketches like distinct counting and most frequent, to compute statistics vital for network analysis.

The culmination of this project is a user-facing dashboard consisting of interactive widgets with charts that empower network administrators with near-real-time insights into network traffic as well as a streamlined data pipeline to efficiently compute statistics. The success of this endeavor positions it for a seamless transition to production after undergoing quality assurance testing.

Worked alongside a research team headed by Professor James Landay to build a fitness tracker that leverages compelling narratives and artwork to encourage positive behavior. Developed and maintained core features on the Android platform to ensure readiness for full release.