Computer Vision Engineer
Walmart Oct 2025 — PresentDallas, TX
- Leading a stealth Proof of Concept for AI-driven theft prevention across Walmart international stores (Chile, Mexico, Canada) — owning end-to-end strategy, research direction, system architecture, and implementation.
- Driving project vision through daily standups, milestone planning, success metric definition, and stakeholder demos to secure funding for production rollout.
- Designing real-time theft detection pipelines on live surveillance feeds — optimizing for suboptimal camera angles, compression artifacts, and variable lighting.
- Architecting action recognition systems using pose estimation, human detection, object tracking, and cart "safe zone" modeling to identify concealment behaviors.
- Implementing state-of-the-art models including YOLOv8, YOLOv11, YOLOv26, SAM2/SAM3, MMAction2, and GroundingDINO.
- Developing temporal reasoning pipelines that correlate pose keypoints, tracked objects, and disappearance events to infer theft scenarios over time.
- Addressing data scarcity by curating in-store footage, adapting public datasets, manually annotating custom datasets, and generating AI-synthesized theft videos.
- Designing human-in-the-loop validation workflows for continuous model refinement and false-positive reduction.