Experience
Research Engineer
- Lead engineer on TuringDB, a vectorised, column-oriented analytical graph database engine written in C++23.
- Led the design and implementation of a custom columnar network protocol, reducing query latency 32x.
- Researched, designed, and implemented novel cache-conscious join algorithms, over 50x faster than competing engines.
- Led the design and implementation of distributed TuringDB using a hybrid S3 and Raft architecture.
- Led the design of a page-table-like index for ultra-low-latency node and edge lookups.
- Cut a client's total memory consumption by 83% by profiling access patterns and moving to mmap-based access.
Software Engineer
- Automated customer subscription management by integrating Stripe webhooks, eliminating occurrences of lapsed subscribers with access to the app.
- Developed a feature to auto-generate barcodes for products based on UPCs and SKUs, streamlining the process for the operations team.
Education
Masters by Research in Computer Science
Thesis: Accelerating the NVMe-oF protocol with Homa
Bachelor of Science in Artificial Intelligence and Computer Science
Thesis: Evaluation of state of the art network stacks
Projects
NVMe over Homa
- Extended Linux NVMe-oF driver to utilize Homa, a datacenter-specific transport protocol.
- Mapped NVMe-oF procedures from TCP's stream abstraction to Homa's message-oriented RPC abstraction.
- Achieved up to 24% (P50) and 26% (P99) latency reduction compared to TCP backend, validated using fio test-suite.
Open Source Contributions
- ntopng (Networking, C++) - submitted PRs to the packet processing software, improving the SYN Flood detection and Elasticsearch logging.
Publications
Designing Transport-Level Encryption for Datacenter Networks
Technical Skills
- Languages: C/C++, TypeScript/JavaScript, Go, SQL (Postgres), Lua, HTML/CSS
- Frameworks: React, Node.js, Redux, ExpressJS, Go-chi, Flask, GraphQL
- Developer Tools: Git, Docker, Kubernetes, Skaffold, Kaniko, Jenkins, Google Cloud Platform, NeoVim, GDB, ftrace, perf, eBPF