Skip to content
View shashankdurgad's full-sized avatar

Highlights

  • Pro

Organizations

@Onflow-AI

Block or report shashankdurgad

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
shashankdurgad/README.md

Shashank Durgad

SWE Intern @ Overmind · CS @ UCL

I build agentic AI systems and the infra that make them measurably better. I also enjoy exploring the frontier of Applied AI, especially in agentic human emulation.


🚀 What I'm working on

Overmind Building an AI Engineering Copilot, which involves agent observability, automated evaluations and model finetuning
Avenir-UX AI research on multimodal agents that simulate real users and generate UI/UX feedback · white paper

🏆 Built at Hackathons/Competitions

Sentinel — 🥇 Environment Track, NVIDIA GTC Hack for Impact → Won Dell GB10

Autonomous agents negotiating satellite collision avoidance in seconds to reduce debris cascades.

  • Dual-path risk classifier — rule-based thresholds + XGBoost (8 features, 4 risk classes), serving a conservative max of both
  • Real-time feature pipeline over live CelesTrak conjunction and NOAA space weather feeds

Python · XGBoost · FastAPI · Nemotron-30B

Sentrix — 🥇 1st, AIEngine UCL BUILD AI Festival → Won free trip to NVIDIA GTC 2026

An agentic police force to monitor autonomous agents in production

  • LangGraph patrol swarm with pheromone-based sampling and quorum voting, feeding a 4-stage investigation pipeline
  • ETL layer ingesting raw agent outputs (A2A messages, emails, PRs, docs) into SQLite, backed by NetworkX for graph analysis

Python · LangGraph · SQLite · NetworkX · FastAPI · SSE

Onflow — 🥇 1st, Bloomsbury Startup Academy Demo Day · 🥈 2nd, Gemini 3 London Hackathon

Multimodal agents that simulate real users to generate UI/UX feedback.

  • LangChain agents driving the simulation, with Playwright for browser automation and visualising agentic navigation
  • GitHub API + Gemini CLI closing the loop with codegen fixes
  • Judged by Google and industry leaders; shortlisted for the Google AI Futures Fund interview

Next.js · LangChain · Playwright · GitHub API · Gemini CLI


🛠️ Tech Stack

Languages

Python TypeScript Swift Java C++ SQL

AI / ML

LangGraph LangChain PyTorch scikit-learn XGBoost pandas NumPy OpenCV MediaPipe

Frameworks & Infra

Next.js React FastAPI Django PostgreSQL SQLite Docker Playwright OpenTelemetry Git


Other cool stuff

  • 🎓 VP, UCL Data Science Society · First Year Rep, UCL AI Society · Foundry Labs Cohort 4

📫 Reach me

Email · LinkedIn · GitHub

Popular repositories Loading

  1. turings-brainchildren turings-brainchildren Public

    Python

  2. admi-Onflow-user-website admi-Onflow-user-website Public

    TypeScript

  3. Algorand-dApp-Quick-Start-Template-TypeScript Algorand-dApp-Quick-Start-Template-TypeScript Public

    Forked from Ganainmtech/Algorand-dApp-Quick-Start-Template-TypeScript

    Kickstart your Algorand dApp in minutes. Full-stack TypeScript template with wallet connect, payments, NFT minting, token creation, and smart contract support

    TypeScript

  4. ESP32-iBeacon-indoor-positioning ESP32-iBeacon-indoor-positioning Public

    Forked from simonbogh/ESP32-iBeacon-indoor-positioning

    Indoor positioning of iBeacon tags (tagged people, dogs, cats and objects) by trilateration.

    JavaScript

  5. ef_hack_c ef_hack_c Public

    Forked from DJRHails/blindsighted

    Swift

  6. functiongemma-hackathon functiongemma-hackathon Public

    Forked from cactus-compute/functiongemma-hackathon

    Getting started repo for the Cactus x DeepMind Hackathon

    Python 1