Designed, led, and implemented numerous customer-facing features as well as other infrastructure, devexp, AI and more across the stack. Work included owning design and implementation of various distributed systems, data processing pipelines, internal web tooling from start to finish working independently and working across teams like data, marketing, customer experience, product, finance, firmware, hardware ops, etc.

Realtime pet ranking system, time-series data processing pipeline with timescaledb, redesigned the customer support experience for troubleshooting Fi collars and user setups from the ground up. Used RAG-based LLM workflows to derive insights from current and historical device telemetry and state to aid in debugging.

Created and owned Fi's internal video labeling platform for training AI models for pet behavior detection.

Creating standardized AI workflows to improve productivity across engineering, product, and other stakeholders.

Tech used highlights: Typescript, React, Next.js, GraphQL, Websockets, Python, Postgres, Timescaledb, Turborepo, Github Actions, Databricks, SQL, Docker, PGLite, AWS, CDK

Briefly worked on engineering foundations team, building infrastructure for bare-metal hardware to facilitate Uber's multi-cloud strategy (AWS, GCP, self-hosted).

Significant contributor to Uber's new observability system for application logs replacing the existing ELK stack. Led SRE/productionization efforts for the new system to improve reliability and reduce maintenance/toil. The system houses petabytes of log data in a NoSQL columnar store from Ubers production systems and is used to diagnose problems and analyze trends.

Key maintainer of data center turnup automation tooling used to bootstrap new datacenters from a collection of unprovisioned servers and cloud instances to accepting production traffic.

Designed, implemented and maintained components within Uber's proprietary service-mesh system implemented in Go, including service discovery, distributed health-checking, and traffic management. Performed load testing and improved safety of operating these systems with integration tests and staging/canary environments.

First engineer of over 50 on a team focused on designing and implementing the in-house customer support system at Uber, helping millions of riders and drivers solve issues and improve their experience with the Uber product. Designed end-to-end distributed systems of Go and Python microservices, APIs (internally-facing and publicly-facing), and databases to accomplish business goals. Balanced designing for the future and making trade-offs to accomplish goals in a timely manner. Mentored junior engineers and established coding standards and a high-performance culture.

Writing software for the computational neuroscience lab.

Developed language translation tools for neuroscience DSL's (eg. from the NCS simulator language to NeuroML, an emerging standard for modeling neural simulations).

Integrated a ROS-based robotics simulation in Gazebo using a Willow Garage PR2 robot to the NCS brain simulator to show decision-based learning in a practical environment.

Built a RESTful Python service to manage users, run brain simulations, view the results of a simulation, and configure real-time data-streaming from the simulator remotely.

Developed a JSON-based modeling language for neuroscientists to write, and configure neural models, and to transfer simulation information between different services, such as between a web-based modeling tool and a model database based on MongoDB.

Developed a documentation system that compiles simple Markdown documents into a Twitter Bootstrap-based online documentation system. It can also be compiled into a PDF for offline documentation.

Advised multiple senior-project groups working with the lab developing neuroscience tools on strategies and provided direction for the projects.

Creating a system for the School of Journalism that crawls news websites in certain areas of the country and use a graph database (neo4j) to see how these sites are interconnected and NLP (natural language processing) to determine the sentiment of articles, and what kind of topics the article covers.

Design and implement an automatic grading system for the CS302 (Data Structures) class. The system is a web-based application (built using Flask) where students upload their assignments, and they are built using GCC/Clang and are then unit-tested to grade them. I also graded handwritten tests and taught a few different classes on expression trees, binary search trees and hash tables.