About SelfInspectionSelfInspection is building a Vehicle Condition Intelligence Platform: the system of record for vehicle condition across a vehicle's entire lifecycle. Our AI turns fragmented inspection and handoff data into a trusted, continuously updated source of truth. Rental operators, fleets, dealerships, finance providers, insurers, transportation companies, automotive marketplaces and mobility platforms use it to manage risk, protect assets and reduce disputes. The roleYou'll own the computer vision behind our next capture experience. It is a video-first inspection that detects damage live on the phone and shows it in AR while the user walks around the vehicle. You'll also own the server-side pipeline that turns that video into an evidence-backed condition report. This is a hands-on senior IC role on a small team. You'll take models from data and labeling through on-device deployment, production monitoring and retraining. You'll set technical direction for vision, mentor engineers, and raise the bar in code and design reviews. You'll work closely with mobile, backend and product engineers to turn customer requirements into shipped features. Full-time, remote. What you'll build
• Live, on-device damage detection. Detection and segmentation models that run on the camera stream on mid-range iOS and Android phones, inside a real-time frame budget.
• AR damage overlays. Track detections across frames and anchor them to the vehicle using ARKit/ARCore pose and depth, so marks stay put as the camera moves.
• Real-time capture checks. Blur, motion, glare,



exposure and coverage checks that guide the user to a complete, usable video, starting in our mobile web capture flow.
• Video-to-evidence pipeline. Best-frame extraction, high-resolution re-detection on keyframes, cross-frame and cross-view de-duplication, and mapping each damage to panel, type and severity.
• 3D vehicle understanding. Camera pose, depth estimation and Structure-from-Motion to place damage on the vehicle and support 360° and 3D views.
• Specialized reads. Tire condition and DOT codes, wheel damage, VIN plates and odometers, feeding structured, evidence-linked data into the condition report.
• The data engine. Labeling guidelines, active learning on hard cases, golden evaluation sets, regression gates and drift monitoring for every model you ship. What you bring
• 5+ years building computer vision systems, including models you trained, shipped and maintained in production.
• At least one model shipped on-device or at the edge under a hard latency budget. Hands-on with Core ML, TensorFlow Lite/LiteRT, ONNX Runtime or TensorRT, plus quantization and profiling on real phones.
• Deep detection and segmentation experience on messy real-world imagery: small objects, reflections,



occlusion and class imbalance.
• Video

experience:
multi-object tracking, temporal smoothing, keyframe selection, and motion and blur estimation.
• Solid multi-view geometry: camera models, pose estimation, homographies and triangulation.
• Strong Python and PyTorch. C++, Swift or Kotlin for performance-critical on-device code.
• Evaluation discipline: golden sets, per-class precision and recall at a chosen operating point, and failure analysis that drives the next labeling round.
• Clear written communication in a remote team. You can tell product and sales what is feasible, by when, and at what accuracy.
• BS/MS in Computer Science, Electrical Engineering or a related field, or equivalent practical experience. Nice to have
• Automotive damage, inspection or insurance-claims experience, or other reflective-surface problems.
• ARKit/ARCore: world tracking, anchors, scene depth, and LiDAR on iPhone Pro models.
• Monocular depth estimation, Structure-from-Motion (COLMAP or similar), NeRF or Gaussian Splatting.
• In-browser inference with ONNX Runtime Web, WebGPU or TensorFlow.js.
• VLMs such as Gemini, GPT, Claude or Qwen-VL for verification and structured extraction, backed by rigorous evals.
• OCR on hard surfaces: tire sidewalls and DOT codes, VIN plates, odometers.
• MLOps: MLflow, model registries, shadow deployments, and GPU inference serving on AWS.
• Perception work in robotics or autonomous systems.

📌 Senior Computer Vision Engineer (México)
🏢 Self Inspection
📍 México