Computer Science student & developer building AI-powered products, full-stack systems and experimental technology.
- Status
- Available
- Local
- Based
- Nagpur, IN
- Focus
- AI · CV · Full-stack
About
- Studying
- B.Tech, Computer Science Engineering
- University
- Ramdeobaba University, Nagpur
- Location
- Nagpur, India · 21.14°N 79.08°E
- Currently
- Working on something new. Stay tuned.
- Also
- Sports Co-Head, CSE Committee
I'm Vineet, a Computer Science Engineering student at Ramdeobaba University, Nagpur. I build full-stack products and AI systems, and I'm most at home where models meet messy, real inputs like a CCTV feed, a voice command or a security log.
Hackathons taught me to ship. Projects taught me to make things hold up. I care about clean architecture, fast interfaces and software that does one thing really well.
- 01
Full-stack
Products from schema to interface: Next.js, Node, Postgres, Supabase.
- 02
AI / ML
Computer vision and LLM systems that run under real constraints.
- 03
Build culture
Hackathons, fast prototypes and leading small teams to a demo.
- 01Computer Vision · Fire safetyMedha Medithon 2026 · Top 5
Turns existing CCTV into a live evacuation system. YOLOv8 spots fire, smoke and people, a hazard grid scores every zone, and A* routes each person to the safest exit.
Key featureHazard-aware A* routing that re-plans the moment fire spreads
- Fire, smoke & occupancy detection
- Live hazard schematic per floor
- Browser camera → FastAPI over WebSockets
- YOLOv8
- OpenCV
- A*
- FastAPI
- WebSockets
- 02Voice AI · Self-learning assistantWinner · ANVESHAN 2026
A voice-first desktop assistant that teaches itself. Ask for something it can't do and it writes the skill, sandbox-tests it, registers it and then does the task, all in one turn. The latest build, NIVA NEXUS, adds a full-screen gesture HUD with a zoomable universe.
Key featureAuto-learn-then-do: new skills written, tested and run on demand
- 40+ skills: apps, browser, WhatsApp, system, research
- Self-improvement loop that patches its own code
- Multi-model brain: Groq, Gemini or local Ollama
- Python
- PyQt5
- LLMs
- Speech
- Automation
- 03Surveillance · Smart India HackathonSIH 2026 · PS 26187
Border-surveillance intelligence that runs on existing IP cameras. It detects, tracks and names behaviours, reads number plates, and scores every alert 0–100 with a reason for each point.
Key featureExplainable risk scoring, backed by 1,067 automated tests
- Detection, tracking & re-ID after occlusion
- Loitering, border-facing & night-movement behaviours
- ANPR with cross-camera plate checks
- Python
- YOLOv8
- OpenCV
- FastAPI
- PaddleOCR
- cipher://soc/localoffline-safeEvent stream
- CRITSuspicious PowerShell spawnhost-07
- HIGHBrute-force on SSH (42 tries)gw-01
- MEDNew admin account createddc-02
- LOWOutbound DNS anomalyhost-12
- HIGHLateral movement via SMBhost-03
- LOWExpired cert on serviceweb-04
- CRITSuspicious PowerShell spawnhost-07
- HIGHBrute-force on SSH (42 tries)gw-01
- MEDNew admin account createddc-02
- LOWOutbound DNS anomalyhost-12
- HIGHLateral movement via SMBhost-03
- LOWExpired cert on serviceweb-04
AI triageLikely credential-stuffing followed by lateral movement. Isolate host-03, rotate keys on gw-01.
✓ Correlated 3 alerts✓ Mapped to MITRE T1021✓ Playbook draftedegress 0 B · model local04Security · Local-first AIIn developmentA local-first AI security operations centre. It ingests logs, correlates events and triages threats on-device, so no telemetry ever leaves the machine.
Key ideaLLM-assisted triage that runs fully offline
- Log ingestion & correlation
- On-device threat triage
- Analyst-style incident summaries
- Python
- Local LLMs
- Security
- SOC
Repository private while in development - 05Product · Membership platform
A full gym platform for Nagpur: a 14-page site plus a member portal and admin panel covering memberships, billing, vitals, referrals and points.
Key featureRole-based member & admin portals on one auth system
- Member management & billing
- Referrals, points & shop
- Health & vitals tracking
- JavaScript
- Firebase Auth
- Firestore
- HTML/CSS
More experiments, hackathon builds and work-in-progress live on GitHub.
github.com/vinuah-dev- Exp. 01
Evacuation simulator
SAFE-X's routing idea, in your browser. Click the plan to start fires; every person re-plans with hazard-aware A*.
- Exp. 02
Talk to Jarvis
Ask by voice or text. It answers and drives this page for you.
- Exp. 03
Command palette
Press ⌘K or Ctrl K anywhere to jump, copy or connect.
“Ideas are cheap. Working demos at 4 a.m. are not.”
Timeline
Education, internship, competitions and leadership. The short version.
- Now· Education
B.Tech, Computer Science Engineering
Ramdeobaba University, Nagpur
Systems, algorithms and ML. Most of the learning happens in the build log.
- Internship· Experience
Software Development Intern
Shaibya Solution
Shipping real features end-to-end inside a working team: reviews, testing and deployment.
- Leadership· Leadership
Sports Co-Head, CSE Committee
Ramdeobaba University
Planning and running department sports events, and coordinating the people behind them.
- Hackathons· Competitions
Hackathons & competitions
ANVESHAN · EFOS SkillUp · Medha Medithon · SIH
A winner's trophy, a national runner-up spot, a Medithon Top 5 and an SIH build, each taken from idea to working demo.
- Projects· Engineering
Computer vision & AI systems
SAFE-X AI · Jarvis · SENTINEL-X · CIPHER
Applied CV and LLM systems built around real constraints such as latency, privacy and safety.
- 01
Languages
- Java
- C++
- 02
Frontend
- Three.js
- 03
Backend
- Node.js
- Express
- Flask
- REST APIs
- 04
AI / ML
- TensorFlow
- TFLite
- HuggingFace
- 05
Database
- PostgreSQL
- MongoDB
- MySQL
- Supabase
- 06
Cloud / DevOps
- AWS
- Google Cloud
- Docker
- GitHub Actions
- Cloudflare
Public activity · last 13 weeks
28 public eventsPrimary languages across repositories
- Python 38%
- JavaScript 29%
- TypeScript 19%
- HTML 10%
- Kotlin 5%
- fire-vdo
SAFE-X AI: browser-camera fire, smoke & occupancy detection with exit routing.
PythonUpdated 22 Jul 2026 - jarvis-ai
Jarvis AI is an award-winning intelligent assistant system designed for voice interaction, desktop automation, and AI-powered task execution. The project was showcased at multiple innovation competitions and received recognition for its practical AI implementation and automation capabilities.
1Updated 9 May 2026 - SIH-TECH-GARUDA
SENTINEL-X: intelligent border surveillance for SIH 2026.
Python 1Updated 17 Sept 2026 - fire-risk-detection-ai
AI-powered real-time fire and smoke risk detection system using computer vision and intelligent hazard analysis.
1Updated 9 May 2026 - Revolution-Gym
Gym site, member portal & admin panel.
HTML 1Updated 24 May 2026 - Fire
SAFE-X AI core: detection, pathfinding and hospital simulation.
PythonUpdated 15 Oct 2026