class Shawon:
def __init__(self):
self.name = "Sudipto Roy Shawon"
self.role = "CSE Undergrad @ North South University"
self.learning = ["Deep Learning", "Computer Vision", "RAG Systems"]
self.stack = ["Python", "PyTorch", "Electron", "Node.js", "React Native"]
self.fuel = ["chai", "Codeforces", "3AM debugging"]
def current_mission(self):
return "Making machines see, think, and snitch on intruders 🕵️"- 🖥️ Shipped LabDesk with IntelligentLab AI — runs a 40-PC lab from one screen
- 🧠 Currently deep-diving into Deep Learning, CNNs & Retrieval-Augmented Generation
- 🏫 Maintaining NSU CSE lab archives so juniors don't suffer like I did
- 📫 Reach me at royshawonsudipto@gmail.com
|
Run an entire computer lab from one screen — self-hosted, no cloud, no per-seat license. Shipped publicly with IntelligentLab AI; installers are up on the releases page and it's already benchmarked against commercial tools like Faronics Insight and Classroom Spy.
🤖 A separate AI proctor layer sits on top of this suite — students emit 10-second telemetry packets (window titles, dwell times, clipboard hashes, never screenshots), filtered through a three-tier funnel that resolves most cases locally and escalates only ambiguous ones to an LLM, so the teacher reads a ranked alert feed instead of watching 40 tiles. Not part of the public release — ask me for a walkthrough. |
|
Watches every syscall on a Linux machine and flags data-stealing processes. Two-layer defense: handwritten rules + an LSTM / Isolation Forest trained on a public intrusion dataset.
|
An autonomous car that dodges obstacles, detects fire in real time, and fires off a Telegram alert the second things get hot.
|
|
Deep CNN pipeline that reads chest X-rays and separates healthy lungs from pneumonia cases.
|
🎬 SceneItMovie discovery app — browse what's trending, search, build a watchlist, and get personalized picks.
|
|
Full-stack task planner that helps students manage workload, track productivity, and optimize study schedules.
|
Ask questions, get answers straight out of your documents — built while learning retrieval-augmented generation.
|
⭐ If any of my repos helped you, a star goes a long way ⭐
