Open to opportunities

wassim sghaier

@wassimsghaier

Message

I build production computer vision systems, from data pipelines to low-latency AWS deployment.

Tunisia

Message

What I'm looking for

I'm looking for computer vision or machine learning work where I can own systems from data and model development through reliable production deployment. I'm open to relocating to Germany, the Netherlands, Belgium, or Luxembourg.

At Bauleiter Günther AI, I replaced manual photo-based cable QC with an instance-segmentation pipeline achieving 95.4% F1 for hazardous-rock sanding verification and 100.0% F1 for cable-overbending detection on held-out data.

I build the full computer vision workflow: benchmarking YOLO and RF-DETR models, creating CVAT and Segment Anything Model annotation pipelines, running active-learning loops, and deploying ONNX-quantized models on serverless AWS at under 120ms per frame.

At Plesmosense, I built an end-to-end Streamlit ML platform that cut new-dataset setup time by 70% and delivered calibrated, uncertainty-aware clinical predictions. I also developed a real-time multi-task face-analysis system handling recognition, emotion, and gender/age inference on live video.

I enjoy turning computer vision and machine learning prototypes into practical systems, from segmentation and detection to LLM fine-tuning and cloud pipelines.

Experience

Work history, roles, and key accomplishments

Computer Vision Engineer

Bauleiter Günther AI

Jan 2026 - Jun 2026 (5 months)

Replaced manual photo-based cable QC with an automated pipeline, achieving 95.4% F1 on hazardous-rock sanding verification and 100.0% F1 on cable overbending detection. Cut per-image annotation time by 40-60% and scaled the labeled set to 3,500+ images using a self-hosted CVAT workflow with active learning.

Computer Vision YOLO CVAT SAM Onnx AWS Active Learning