- Designed BigQuery and Avro schemas for Twitter’s next generation offline feature store in GCP; implemented and scheduled set of Dataflow jobs for loading on-prem Twitter data to BigQuery - Contributed performance improvements to feature store online client to enable high cardinality fetches at low tail latencies (4k keys at p99 of < 300ms)- Designed Scala DSL for Twitter’s on-prem feature store solution; implemented tooling for batch generation of feature data. Acted as interim tech lead while primary was on leave.- Designed and implemented backend for new privacy controls; provided on-call support and bug fixes- Implemented improvements to TellApart data warehouse, including Hive UDFs and custom identity graph Hive tables

Built a more scalable backend for a broadcast messaging feature.New backend can support 3-4x more recipients than the original implementation.

COMPSCI 100: Introductory undergraduate course on data structures and algorithms in JavaECE 451: Introductory undergraduate computer engineering course on basic computer architecture ECE 590: Introductory Masters' computer engineering course on digital circuits and basic computer architecture.

Worked on fault tolerant, highly available solution for Hadoop. Fixed critical customer-facing bugs, created distributed installation and configuration scripts, and improved product stability.

Designed a normalized relational database schema for storing patent information.Wrote multithreaded Python scripts to process bulk patent XML files and write their contents to a MySQL database.Average process and insertion time for a ~500 MB patent XML file was 110 seconds.

Programmed gambling behavior research experiments in MATLAB using Psychophysics Toolbox SDK. Ran subjects for data collection on Tobii eye tracking systems.

Designed and implemented modular vehicle operator console using Netbeans Platform and third party Java libraries. Modular design allows for extension by enduser and for multiple vehicle support.