Computer Vision & Visual Perception Lead

Labor Mexicana, SC; Saltillo; Presencial; MXN 800.000 - 1.100.000; Jornada completa

Descripción de la vacante

Labor Mexicana invites a Founding Computer Vision & Visual Perception Lead to architect and implement a proprietary industrial vision system. You will transform images and video into reliable, explainable inspection results under real industrial conditions.

The role emphasizes building robust algorithms for detection, segmentation, and tracking, while ensuring cross-device compatibility and offline capability in an industrial setting.

Formación

• Strong programming experience in Python and C++.
• Practical mastery of OpenCV.
• Experience with PyTorch, TensorFlow, JAX or equivalent frameworks.
• Strong knowledge of object detection, segmentation, visual representations and video processing.
• Experience with tracking, optical flow, feature matching, homographies, camera calibration or pose estimation.
• Ability to train, evaluate, optimize and deploy proprietary models.
• Experience converting research prototypes into robust software.
• Knowledge of ONNX, TensorRT, OpenVINO, CUDA or other optimization and inference tools.
• Ability to work with industrial cameras, webcams, mobile devices and other image sources without depending on a single brand.
• Ability to evaluate escapes, false rejects, temporal stability, latency and visual coverage—not only accuracy or AUROC.

Responsabilidades

• Develop a visual perception engine capable of analyzing image and video sequences under changing conditions.
• Build algorithms for defect detection, segmentation and localization.
• Implement video processing and automatic selection of useful frames.
• Develop tracking and temporal association of detections.
• Handle motion compensation, perspective changes and scale variations.
• Perform image registration and cross-view correspondence.
• Provide visual coverage estimation and identify unobserved areas.
• Deduplicate findings across frames and detect invalid images.
• Create offline/in-edge inference across devices and architectures.
• Deliver a reusable architecture for different parts and devices.

Conocimientos

Python

C++

OpenCV

PyTorch

TensorFlow

JAX

Object detection

Segmentation

Video processing

Tracking

Optical flow

Feature matching

Homographies

Camera calibration

Pose estimation

ONNX

TensorRT

OpenVINO

CUDA

Industrial cameras

Herramientas

OpenCV

PyTorch

TensorFlow

JAX

ONNX

TensorRT

OpenVINO

CUDA

Descripción del empleo

Founding Computer Vision & Visual Perception Lead

Labor Mexicana is looking for the technical leader who will develop a new proprietary industrial visual inspection technology.

This is not a position for installing or configuring commercial vision systems. We are looking for someone who can build the vision algorithms: transforming images and video from different devices into reliable, explainable and traceable inspection results under real industrial conditions.

The Technical Challenge

Develop a visual perception engine capable of analyzing image and video sequences under changing position, perspective, distance, lighting and motion.

The solution must identify defects and anomalies, determine which areas have already been observed, avoid counting the same finding multiple times and recognize when the available visual evidence is insufficient to produce a reliable result.

The technology must remain independent of any specific camera manufacturer and be adaptable to different image-capture and computing devices.

What You Will Build

• Algorithms for defect detection, segmentation and localization.
• Video processing and automatic selection of useful frames.
• Tracking and temporal association of detections.
• Compensation for motion, perspective changes and scale variations.
• Image registration and correspondence across multiple views of the same object or surface.
• Visual coverage estimation and identification of unobserved areas.
• Deduplication of findings detected across different frames.
• Detection of invalid images caused by blur, poor lighting, obstruction or out-of-distribution conditions.
• Anomaly-detection models for applications with limited defect samples.
• Local and offline inference across different computing platforms.
• A reusable architecture that can support different parts, processes and image-capture devices.
• Visual evidence and metrics that explain why the system produced each result.

Required Profile

• Strong programming experience in Python and C++.
• Practical mastery of OpenCV.
• Experience with PyTorch, TensorFlow, JAX or equivalent frameworks.
• Strong knowledge of object detection, segmentation, visual representations and video processing.
• Experience with tracking, optical flow, feature matching, homographies, camera calibration or pose estimation.
• Ability to train, evaluate, optimize and deploy proprietary models.
• Experience converting research prototypes into robust software.
• Knowledge of ONNX, TensorRT, OpenVINO, CUDA or other optimization and inference tools.
• Ability to work with industrial cameras, webcams, mobile devices and other image sources without depending on a single brand.
• Ability to evaluate escapes, false rejects, temporal stability, latency and visual coverage—not only accuracy or AUROC.

Highly Valuable Experience

• Industrial visual inspection.
• Robotics, robotic perception, SLAM or 3D reconstruction.
• Real-time video analytics.
• Anomaly detection and learning with limited defect samples.
• Self-supervised learning, visual foundation models or domain adaptation.
• Edge deployment and fully offline operation.
• Development of vision products used by operators in real working environments.

This Position Is Not a Fit If Your Experience Is Primarily Limited To

• Installing cameras or sensors.
• Configuring prebuilt tools from a single vision-system manufacturer.
• Programming PLCs without developing computer-vision algorithms.
• Training classifiers on clean image datasets without deploying them under real operating conditions.
• Creating AI demonstrations that have not been evaluated with video, motion or real-world process variations.

The Position

This is a founding technical leadership position with direct responsibility for the product’s architecture, algorithms and technical evolution.

You will collaborate with automation, software and systems-integration specialists, while maintaining primary ownership of the proprietary visual-intelligence technology.

The ideal candidate combines deep technical ability with the practical judgment required to transform an algorithm into a reliable, repeatable and commercially viable industrial product.

Location

Saltillo Mexico, with direct collaboration with the Labor Mexicana team and occasional travel to industrial facilities for testing and validation.