Skip to content
View dhyeydaftary's full-sized avatar
8 followers · 4 following

Achievements

Achievement: Pair ExtraordinaireAchievement: Pull Sharkx3Achievement: YOLOAchievement: Quickdraw

Achievements

Achievement: Pair ExtraordinaireAchievement: Pull Sharkx3Achievement: YOLOAchievement: Quickdraw

Block or report dhyeydaftary

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
dhyeydaftary/README.md

Dhyey Daftary

Color dot-matrix portrait of Dhyey Daftary, revealing on load.

Typing SVG

LinkedIn Email GitHub

~/ whoami

Early-career engineer, still figuring out where full-stack ends and AI begins. Based in Ahmedabad, India. I write code that mostly works, and bugs that look intentional.

~/ currently building

RSNA Knee Abnormality Detection — Kaggle competition, predicting 12 clinically important knee abnormality findings from multimodal MRI (image and radiology report text at train time, image-only at inference). The real problem isn't the modeling — it's that only 58 of 4,407 training studies carry direct labels; the rest have to be handled as a weak-supervision problem built on noisy report-derived signal. Currently deep in data verification: confirmed patient-level leakage isn't a risk (4,407 unique patients, one study each, verified against DICOM headers), resolved laterality coverage across manufacturers, and now designing a validation strategy that can be trusted with a 58-study labeled core. No model trained yet — getting the data and validation right first.

Python · PyTorch · pydicom

github.com/dhyeydaftary/rsna-knee-abnormality-detection

~/ toolbox

Languages
Languages


Frontend
Frontend
Jinja2


Backend & APIs
Backend & APIs
Django REST Framework


Data & Databases
Data & Databases
SQLAlchemy


Real-time
Socket.IO · WebSockets


AI / ML
AI / ML
CatBoost · SHAP · Explainable ML


Web3
Web3
Arbitrum · Stylus


Tools & Platforms
Tools & Platforms

~/ engineering map

Engineering map: an isometric cross-section of Frontend, Backend, Data, and AI/ML layers, with a Web3 branch. Amber paths trace each project's real coverage — campus-connect spans Frontend through Data; womens-safety-alert-system spans Backend through Data; AMR-Insight spans Backend through Data through AI/ML; originchain spans Backend through Data, bridging into Web3.

~/ selected work

campus-connect
Full-stack social platform for university communities — 90+ REST endpoints across 10 Flask blueprints, real-time chat over WebSockets, an 18-table PostgreSQL schema, 333 automated tests at 83% coverage.
Flask · PostgreSQL · Tailwind/Jinja2 · Socket.IO
Creator, Lead Developer & Architect — with Urva Shah (design) & Twisha Agrawal
github.com/dhyeydaftary/campus-connect
womens-safety-alert-system
Java-based real-time emergency response system — three-tier RBAC, FIFO alert dispatching, geospatial nearest-responder assignment, multithreaded background monitoring, MySQL/JDBC persistence. Singleton, Factory, and Observer patterns throughout.
Java · MySQL · JDBC
Solo — Semester 2 coursework
github.com/dhyeydaftary/womens-safety-alert-system
AMR-Insight
Explainable ML platform predicting antibiotic resistance across 15 antibiotics — 15 CatBoost models with native SHAP explainability, WHO AWaRe classification, and Gemini-assisted lab-report extraction. 98 automated tests, a 13-phase security hardening pass.
Django/DRF · CatBoost · Node/Express · MongoDB
ML backend & gateway; led security hardening — with Urva Shah & Ansh Patel (frontend)
github.com/dhyeydaftary/antibiotic-resistance-intelligence-platform
originchain
Decentralized creator-identity and proof-of-origin platform on Arbitrum — client-side content hashing, IPFS pinning, on-chain registration via Rust/Stylus smart contracts, wallet-based auth via Sign-In With Ethereum.
Express · Prisma/PostgreSQL · Rust/Arbitrum Stylus · Next.js
Backend & smart contracts — built with Nandish Patel & Frenil Patel
github.com/originchain-labs/originchain · Live

~/ contribution activity

GitHub contribution activity over the past year, shown as an isometric calendar.

Pinned Loading

  1. antibiotic-resistance-intelligence-platform antibiotic-resistance-intelligence-platform Public

    AMR-Insight is a research and education platform exploring whether patient and lab data can support explainable, evidence-grounded antibiotic-resistance prediction, a transparent, SHAP-interpretabl…

    JavaScript 2

  2. originchain-labs/originchain originchain-labs/originchain Public

    Decentralized creator identity & proof-of-origin platform on Arbitrum

    TypeScript 1

  3. campus-connect campus-connect Public

    Campus Connect is a student networking platform built with Flask and PostgreSQL. Students can build profiles, discover and register for events, follow real-time campus activity, and connect with pe…

    HTML 1

  4. womens-safety-alert-system womens-safety-alert-system Public

    Java-based real-time emergency response system with 3-tier RBAC, FIFO alert dispatching, geospatial nearest-responder assignment, multithreaded background monitoring, and MySQL/JDBC persistence. Im…

    Java