Sr. Computer Vision Engineer
Panoptyc2 days ago
Location
Remote
Workplace
Remote
Type
Full Time
Salary
USD 160,000 – 220,000
Level
Senior
Role
Computer Vision Engineer
Posted
Jul 23, 2026
Full TimeRemoteSenior
The role
Summary
Senior Computer Vision Engineer at Panoptyc, a supply chain visibility platform company, responsible for architecting and deploying cutting-edge computer vision models for retail object recognition and edge deployment. This role requires deep expertise in custom YOLO model training, vision-language model (VLM) fine-tuning, and edge optimization for real-world retail applications, combined with strong software engineering fundamentals and technical leadership capabilities.
What you'll do
Custom Object Detection Model Development: Design, architect, and train custom computer vision models optimized for retail object recognition including inventory items, shelf arrangements, and product placements; establish evaluation frameworks specific to retail scenarios; iterate on model architectures based on production performance and business requirements
Vision-Language Model Integration and Fine-tuning: Evaluate, fine-tune, and deploy open-source vision-language models (LLaVA, Qwen-VL, InternVL, PaliGemma) for advanced product understanding and scene reasoning; build and optimize vision-language-action (VLA) pipelines that translate visual understanding into downstream business decisions
Edge Model Optimization and Deployment: Apply advanced optimization techniques including model quantization, pruning, architectural optimization, and knowledge distillation to deploy state-of-the-art models on edge devices with latency and memory constraints; ensure models run efficiently on NVIDIA Jetson and similar constrained hardware
Production ML Infrastructure Development: Build robust, production-grade ML systems that move beyond prototypes; establish CI/CD pipelines for model deployment, implement model monitoring systems, create data validation frameworks, and maintain comprehensive documentation for reproducible research
Dataset Engineering and Annotation Workflows: Design and implement scalable data pipelines and annotation workflows for continuous model improvement; manage diverse retail datasets, implement data versioning, track dataset lineage, and ensure high-quality training data across various retail scenarios and edge cases
Research and Innovation Leadership: Stay current with latest computer vision and multimodal AI research; prototype emerging architectures and techniques; critically evaluate research papers to determine production viability; balance innovation with stability in production systems
Technical Mentorship and Best Practices: Mentor junior and mid-level computer vision engineers; establish best practices for model development workflows, code quality standards, and experimentation protocols; conduct technical reviews of model implementations and provide guidance on architectural decisions
Cross-functional Collaboration: Work seamlessly with hardware engineers on device-specific optimizations, collaborate with full-stack engineers on inference API design, and partner with product team to align technical capabilities with business requirements
Model Performance Optimization: Conduct comprehensive performance analysis of deployed models; identify bottlenecks in inference pipelines; optimize end-to-end latency from image capture through model inference; balance accuracy requirements against real-time performance constraints
Experiment Tracking and Reproducibility: Maintain rigorous experiment tracking using MLflow or Weights & Biases; document model hyperparameters, training conditions, and performance metrics; ensure all results are reproducible and enable collaboration across the team on model development
What we look for
Technical
Object Detection Model DevelopmentAbility to design custom YOLO architectures tailored to retail environments, handle class imbalance, manage multi-scale object detection challenges, and optimize for real-time inference performance
Model Fine-tuning and Transfer LearningExpertise in adapting pre-trained models to domain-specific tasks, understanding data distribution shifts, and implementing effective transfer learning strategies for retail recognition tasks
Model Quantization TechniquesProficiency with post-training quantization (PTQ) and quantization-aware training (QAT) methods to reduce model sizes and latency without significant accuracy loss
Distributed TrainingExperience scaling model training across multiple GPUs or TPUs, understanding gradient synchronization, and optimizing training pipelines for efficiency
Model Evaluation and MetricsDeep understanding of computer vision evaluation metrics (mAP, precision, recall, F1 score), ability to design appropriate evaluation protocols for retail use cases, and interpret results critically
Data Pipeline EngineeringCapability to build robust data ingestion, preprocessing, and augmentation pipelines that handle edge cases and scale to production data volumes
API Design and IntegrationExperience designing RESTful or gRPC APIs for model inference, handling batch requests, managing latency requirements, and integrating models into larger systems
Education
Bachelor's Degree in Computer Science, Computer Engineering, or Related FieldStrong foundational knowledge in algorithms, data structures, linear algebra, and mathematics required for understanding and implementing machine learning systems
Master's Degree or PhD in Computer Vision, Machine Learning, or Related Discipline (Optional)Advanced degree demonstrates deeper theoretical understanding of computer vision fundamentals, multimodal learning, and research-oriented approach to problem-solving
Experience
4+ Years Production Computer Vision ExperienceDemonstrated track record of shipping computer vision models to production environments; experience iterating on models based on real-world performance data and user feedback
2+ Years Deep Learning Framework DevelopmentSubstantial hands-on experience with PyTorch or TensorFlow for implementing custom architectures, training complex models, and debugging deep learning systems
2+ Years Edge ML DeploymentProven experience deploying machine learning models to edge devices with constraints on compute, memory, and power; understanding of hardware-software co-design
1+ Years Technical Leadership or MentorshipExperience mentoring junior engineers, establishing technical best practices, conducting code reviews, and driving technical decisions on engineering teams
Skills
Required skills
Computer Vision Engineering4+ years of hands-on production computer vision experience with demonstrated track record of shipping models to real-world applications; deep understanding of image processing pipelines, model training, evaluation metrics, and debugging visual models
YOLO Architecture ExpertiseComprehensive mastery of YOLO and YOLO-E object detection models including custom training, hyperparameter tuning, architecture modifications, and performance optimization; ability to diagnose and resolve training convergence issues
Vision-Language Models (VLMs)Hands-on experience fine-tuning and deploying open-source VLMs (LLaVA, Qwen-VL, InternVL, PaliGemma); expertise in model evaluation, prompt engineering, and production deployment at scale
Edge Deployment OptimizationProduction experience optimizing models for edge devices through quantization, pruning, and architectural modifications; proficiency with TensorRT, ONNX Runtime, and deployment frameworks for resource-constrained environments
Vision-Language-Action (VLA) SystemsExperience building and deploying VLA pipelines that translate visual understanding into actionable downstream decisions; understanding of multimodal reasoning and end-to-end perception systems
Software Engineering FundamentalsStrong proficiency in clean code principles, version control (Git), CI/CD pipelines, code review processes, and ability to build maintainable, well-tested ML systems that scale
Production ML SystemsDeep understanding of the gap between prototype notebooks and production systems; experience with model monitoring, data validation, feature pipelines, and handling model drift in real-world deployments
PyTorch FrameworkAdvanced proficiency with PyTorch for model development, training loop optimization, distributed training, and integration with production serving frameworks
Nice to have
NVIDIA Jetson DevelopmentProven experience deploying optimized computer vision models to NVIDIA Jetson family hardware; understanding of Jetson-specific constraints and optimization strategies
Retail or Inventory Management DomainBackground developing computer vision solutions for retail applications such as inventory tracking, product recognition, shelf monitoring, or supply chain visibility; understanding of retail-specific challenges and workflows
VLM Inference OptimizationExperience optimizing VLM inference using tools like vLLM, llama.cpp, or SGLang for high-throughput serving; knowledge of batching strategies, KV cache management, and distributed inference
Synthetic Data and AugmentationExperience generating synthetic training data and implementing data augmentation pipelines to improve model robustness across diverse retail scenarios and environmental conditions
Model Experiment TrackingProficiency with experiment tracking platforms (Weights & Biases, MLflow) for managing model versions, comparing training runs, and enabling reproducible research workflows
AWS ML ServicesExperience with AWS ecosystem including EC2, ECS, Fargate, S3, SageMaker, and Bedrock for building scalable ML infrastructure and production deployment
Research and PublicationsPublished research papers or significant open-source contributions in computer vision, multimodal AI, or edge ML demonstrating thought leadership and commitment to advancing the field
Modern ML Training FrameworksExperience with advanced training frameworks like Transformers library, LitGPT, Unsloth, and other tools that accelerate model development and fine-tuning workflows
Compensation & benefits
Salary
USD 160,000 – 220,000 (annual)
Stock options
Available
Benefits
Remote Work Flexibility
Fully remote position enabling work from anywhere with flexible schedule; eliminates commute time and allows for optimal work environment setup
Cutting-Edge Technology Stack
Work with state-of-the-art computer vision and multimodal AI technologies including YOLO architectures, vision-language models, and edge deployment frameworks; regular exposure to emerging research and innovation
Technical Leadership Opportunities
Significant mentorship and technical leadership responsibilities; opportunity to shape team direction, establish best practices, and influence architectural decisions for computer vision infrastructure
High-Impact Business Applications
Develop models that directly impact retail supply chain visibility and inventory management; see real-world applications of computer vision technology across enterprise customers
Collaborative Engineering Culture
Work alongside talented hardware engineers, full-stack developers, and computer vision specialists in a collaborative environment that values technical excellence and innovation
Professional Development
Access to continuous learning opportunities in rapidly evolving computer vision field; time and resources for exploring emerging research and implementing innovative techniques
Apply for this position
You'll be redirected to the company's application page
More Jobs at Panoptyc
1 other open position
Panoptyc
View all jobs
Panoptyc is a software company that provides supply chain visibility and intelligence solutions for enterprises. The company offers a platform designed to help organizations monitor, track, and optimize their supply chain operations in real-time. Panoptyc's technology enables businesses to gain comprehensive visibility across their supply networks, identify risks, and make data-driven decisions to improve efficiency and resilience. The platform serves companies across various industries that rely on complex global supply chains.
panoptyc.com
Tech Stack
Languages
Python
Frameworks
PyTorchYOLO (YOLOv8, YOLO-E)LLaVA, Qwen-VL, InternVL, PaliGemmaTransformers LibraryvLLM
Databases
S3
Tools
TensorRTONNX RuntimeDockerKubernetesWeights & Biases / MLflowGit / Version ControlNVIDIA Jetson
Other
AWS Services (EC2, ECS, Fargate, Bedrock, SageMaker)Model Quantization and PruningSynthetic Data GenerationCI/CD for ML