I'm a freelance Computer Vision & AI engineer with over 12 years of hands-on experience architecting and delivering 30+ vision-driven solutions across industrial, medical, and embedded domains. Specialized in hyperspectral imaging, real-time AI pipelines, 3D vision, computer graphics, rendering, synthetic data, and custom hardware integration. I am a former technical lead in AI and computer graphics as well as a very experienced technical interviewer.
Impactful computer vision deployments across sectors
Praline classification on conveyor belts for live demonstration purposes
Bean fermentation and quality segmentation via hyperspectral imaging
Detecting foreign material in salad wash lines
Analyzing salad mix proportions using 3D scanning and synthetic mix creation
Real-time mice detection on produce washline
Improving results of realtime megapixel resolution time of flight camera using classical computer vision filtering
Ceiling light detection to infer forklift location and movements
Accurate rackpole detection on low quality highly compressed images
Create a benchmarking procedure to compare multiple existing ANPR systems
Capturing images of laser beams on Mont Blanc and simulating the visibility of different laser intensities and distances.
6-camera real-time person detection on Jetson Nano in security context
Matching broken objects using text recovered from the images
Matching broken complex mechanical components using visual features
Human detection on crane to aid conductor replacing rails. Real time running on CPU.
Human detection on forklifts. Real time running on Jetson Nano
Automated checking if personal protective equipment is worn
Automatically detect when dangerous activities are taking place near railroad crossings
Gaze tracking on budget hardware
Surface flaw classification on steel plates
Automated segmentation of concrete damage and apply business logic
Automated chamfer tolerance and angle check of railway rails using 3D pointclouds
Detect defects in magnets via magnetic field (MagCam)
Creating systems for classification of weaving and knitting machine issues with limited dataset
Hyperspectral sorting of paper
Automated wood flake segmentation using 3D scan
Segmenting overlapping/crossing tree branches pre-chipping to check for length
UE5-based photorealistic fabric rendering
Real time motion capture and inverse kinematic filtering. Rendered using Blender
Vision systems for helping with volleyball training. Player & ball tracking with simple logic
Vision systems for automated trash sorting
Automatic mistake detection during surgeon training for faster grading.
Create a deep learning model that can compress any 3D scan of a face and can generate random faces.
Combining AI research with real-world deployment.
Real-time detection on Jetson Nano and other constrained edge devices.
Custom networks for classification, segmentation, anomaly detection, object tracking and more.
Experienced in 3D reconstruction, synthetic data generation, physics simulations, digital twinning using Blender & Unreal Engine.
Classification of food and paper using spectral-spatial models.
3D-printed mounts, lighting setups, electronics design and custom optics for CV pipelines.
Want a fast, robust CV/AI pipeline? Let’s talk.