I'm Michael Khuri, founding full-stack engineer at Star4ce on a freelance project basis and an M.S. Artificial Intelligence student at San José State University, specializing in autonomous systems.
I build and deploy software, including a multi-tenant dealership retention platform using Next.js, Flask, and PostgreSQL, and a production pump-quoting tool for AcuFlow. At TRSTXCYBER, I integrated client onboarding tools to support sales qualification.
My computer vision projects include training and comparing road-scene detectors on BDD100K and implementing panorama stitching and integration for a Waymo multi-camera team project. Our evaluation found that LaMa inpainting did not restore YOLO detections in masked fields of view.
I'm interested in full-time software engineering opportunities, with a particular focus on computer vision, perception, and autonomous systems.
Always building. Always learning. Autonomy is the future.
- Email: michaelkhuri@gmail.com
🏊 Swimming • 🥾 Hiking •
| Project | Tech | Description |
|---|---|---|
| RoadAware | Python, PyTorch, TorchVision, YOLOv8, OpenCV | Road-scene object detection system using BDD100K dashcam images. Compared YOLOv8m and Faster R-CNN on 10,000 images across 8 traffic classes. Best YOLOv8m: 0.515 mAP@0.5, 0.734 F1, 64.9 FPS. Forked from amarahatta/Object-Detection. Demo + Video |
| Gap Reconstruction for Autonomous Vision | Python, OpenCV, PyTorch, LaMa, YOLO | Implemented panorama stitching and integration for a Waymo multi-camera team project. Team evaluation found that LaMa inpainting did not restore YOLO detections in masked fields of view. |
| EcoRoad AI | Python, Flask, PyTorch, YOLO, OpenCV | Dashcam video analysis pipeline for eco scoring, CO₂ estimate, driving risk, and object-aware safety tips. Built at SF Hacks 2026 in San Francisco. Demo Video |
| Star4ce (Professional Work) | Next.js, Tailwind CSS, Flask, PostgreSQL | Multi-tenant SaaS platform with authentication, analytics dashboards, worker insights, reporting, and subscription payments. |
| NeuronBook | Next.js, Sanity, Flask, React PDF, Tailwind CSS | AI-powered active-reading platform with PDF viewing, Socratic question generation, knowledge graph visualization, and Sanity-backed library data. |
| SmartTuffy | TypeScript, Next.js, Tailwind CSS, OpenAI API | AI assistant that helps students find course information and campus resources through chat-based recommendations. Built in under 24 hours during a hackathon. |
| Acuflow Quote Generator (Professional Work) | Python, Flask, MySQL, JavaScript | Automated quote generation system with PDF creation, email workflows, custom pricing logic, and REST API integration for a manufacturing company. |
| Detection VR | Unity, C# | Immersive VR FPS with environmental scanning, physics-based interactions, modular weapon systems, and responsive VR gameplay. Demo Video |
⭐️ Always learning. Always building. Autonomy is the future.



