Skip to content
View Deva-1903's full-sized avatar
🎯
Focusing
🎯
Focusing
11 followers · 40 following

Achievements

Achievement: Pair ExtraordinaireAchievement: Pull Sharkx3Achievement: YOLOAchievement: Quickdraw

Achievements

Achievement: Pair ExtraordinaireAchievement: Pull Sharkx3Achievement: YOLOAchievement: Quickdraw

Highlights

  • Pro

Block or report Deva-1903

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Deva-1903/README.md

Hi, I'm Deva 👋

MS CS @ UMass Amherst (May 2027) · Backend & ML Systems

I build production backends and measure everything.

Now

  • 🏥 Software Engineering Intern @ TherAlign Health - SMART on FHIR clinical decision support in the EHR prescribing flow (Next.js, Firebase/GCP)
  • 🧠 Culture & Morality Lab, UMass - sole engineer on a self-serve NLP text-analysis platform for researchers (FastAPI, sentence-transformers, React)

Before

  • ⚙️ Backend Engineer @ Wysa (2023–2025) - multi-tenant NHS eTriage backend serving 8+ UK clinical clients and 10,000+ monthly triage submissions; owned security (AWS KMS, 20+ VAPT fixes) and zero-downtime releases
  • 📄 First-author IEEE paper (CONIT 2023) on transfer learning for Alzheimer's classification

Things I've built

Project What it is
provisioning-platform Server-provisioning control plane in Go (REST + gRPC, idempotent workflows) - ~19,700 provisions/min across a 2,500+ server simulated fleet
gpu-scheduler Multi-tenant GPU/CPU scheduler in Go - weighted Dominant Resource Fairness + bin-packing, 542 jobs/min on a simulated 32-GPU fleet
neon-quantized-gemm INT8/INT4 GEMM kernels in C++ for ARM NEON - 394 GFLOP/s, 132× over the naive baseline
ciir_agentic_search Provenance-first agentic entity discovery (Python/FastAPI, 150+ tests) - live demo
Spark-ETL-Optimization Cut Apache Spark ETL runtime ~25% on a ~1.4B-row NYC TLC dataset

Stack I reach for: Python · Go · TypeScript/Node.js · C++ · React/Next.js · PostgreSQL · MongoDB · Redis · Spark · AWS/GCP

Let's Connect

Pinned Loading

  1. gsoc-good-first-issues gsoc-good-first-issues Public

    Forked from hkirat/gsoc-good-first-issues

    Good first issues for GSoC 2023

    1

  2. Whisper Whisper Public

    Forked from Dun-sin/Whisper

    A Fun Application to have a random chat with people safely

    JavaScript 1

  3. dub dub Public

    Forked from dubinc/dub

    An open-source link shortener with built-in analytics + free custom domains.

    TypeScript

  4. LinkFree LinkFree Public

    Forked from EddieHubCommunity/BioDrop

    Connect to your audience with a single link. Showcase the content you create and your projects in one place. Make it easier for people to find, follow and subscribe.

    JavaScript