I work where vision models meet shipped software — training detectors, classifiers, and recognition pipelines, then wrapping them in interfaces people can actually use. Most days I'm somewhere between a Jupyter notebook and a Next.js route.
📖 Deep Dive: My Computer Vision & ML Workflow
Here is how I typically train and deploy computer vision pipelines:
- Data Prep & Processing: OpenCV and Roboflow for robust augmentations, data labeling, and camera stream handling.
- Model Training: PyTorch, YOLOv8, and Vision Transformers (ViT) in local/cloud Jupyter notebooks.
- Model Deployment: Deploying models as optimized Flask/Python API endpoints supporting live RTSP camera feeds.
🛠️ Deep Dive: My Full-Stack Architecture
How I build modern, performant web applications:
- Frontend Stack: Next.js (App Router), React, and Tailwind CSS for designing animation-rich, visual dashboards.
- Type Safety & Logic: End-to-end TypeScript for robust data flow, clean custom hooks, and state management.
- Backend Stack: Node.js, Express, and Python servers to handle authentication, database storage, and model queries.
- akruit_web · private — recruitment platform front-end, the production cut of the public
Akruitprototype. - omniget · private — TypeScript media fetcher; batch downloads and a clean UI over
yt-dlp. - hospital_management_system · private — full-stack HMS in TypeScript.
- bootsnbooties · private — e-commerce store, TypeScript stack.
| Project | What it is |
|---|---|
| argus | AI-powered intruder detection — YOLO detector + ViT crime classifier + live streaming + chatbot. |
| smart-parking-system- | CV + Tesseract OCR for plate reading and slot allocation, with a Flask dashboard over live video. |
| facial-recognition | Recognition pipeline from capture to inference. |
| resume | TypeScript resume site — the design system I iterate on. |
| distance_estimator | Monocular distance estimation experiments. |
| Option | Preference |
|---|---|
| ☕ Fuel | Cold Brew Coffee / Chai |
| ⌨️ Keyboard | Razer BlackWidow V3 TKL |
| 🖱️ Mouse | Razer Basilisk V3 |
| 🎨 Theme | Midnight Dark / Catppuccin Mocha |
| 🪟 OS | Windows (WSL2) / Ubuntu |
| 🧑💻 Editor | VS Code / Zed |
| 🧠 AI assistant | Claude / Codex |
