Lead AI Engineer

We are looking for a computer vision-focused deep learning engineer or research scientist with strong hands-on expertise in developing and training vision models. The ideal candidate should have a solid foundation in image processing, visual recognition, segmentation, defect detection, and representation learning, along with experience in generative AI, diffusion models, or vision-language models. Candidates with primarily NLP, LLM application development, or text-based generative AI experience, but limited hands-on computer vision experience, would not be the primary target for this role.

Responsibilities:

• Computer Vision and Deep Learning: Research, design, and develop advanced computer vision models for image understanding, visual inspection, surface defect detection, segmentation, object detection, and pixel-level scene understanding.
• Generative AI and Diffusion Models: Research and innovate diffusion-based generative models for photorealistic wall-surface simulation, defect synthesis, synthetic image generation, and domain adaptation.
• Vision-Language Models (VLMs): Architect, train, and fine-tune multimodal models and vision-language alignment objectives to connect textual work orders, CAD plans, images, and sensor data for improved visual understanding.
• Large-Scale Data and Auto-Annotation: Build scalable data-centric AI pipelines using active learning, self-training, pseudo-labeling, weak supervision, and synthetic data generation to annotate millions of images, video frames, and point clouds with minimal human intervention.
• Model Optimization and Edge Deployment: Optimize and compress computer vision and deep learning models using INT8 quantization, LoRA, pruning, and knowledge distillation for deployment on Jetson-class edge devices.
• Robotics Perception: Integrate vision models into robotics perception pipelines using ROS 2, enabling real-time visual processing and deployment in robotic inspection and automation systems.
• Research and Production Ownership: Own the complete ML lifecycle, including problem definition, literature review, experimentation, prototyping, model training, offline and online evaluation, optimization, and production handover to perception and controls teams.
• Performance Optimization: Improve model accuracy, robustness, inference latency, and generalization across varying lighting conditions, surface textures, camera configurations, and real-world environments.
• Research and Technical Leadership: Publish internal technical reports and external research papers, stay updated on advances in computer vision and generative AI, and mentor interns and junior engineers.

Requirements:

• Experience: 3+ years in computer vision, deep learning research and development, or applied AI; alternatively, an M. S. or Ph. D. in computer science, electrical engineering, robotics, or a related field, with a strong research or publication record.
• Core Computer Vision Expertise: Strong hands-on experience in image processing, image classification, object detection, semantic and instance segmentation, visual feature extraction, image synthesis, and visual representation learning.
• Deep Learning Frameworks: Advanced proficiency in Python and PyTorch or JAX, with experience training, fine-tuning, evaluating, and deploying deep learning models.
• Generative Models: Demonstrated expertise in diffusion models such as DDPM, Latent Diffusion Models (LDM), and ControlNet, particularly for image generation, defect synthesis, image-to-image translation, or domain adaptation.
• Vision-Language Models: Experience with multimodal transformers and vision-language architectures such as CLIP, DINO/DINOv2, LLaVA, or Flamingo, including vision-language alignment and multimodal representation learning.
• Data-Centric AI: Proven experience developing large-scale computer vision datasets and annotation workflows using active learning, pseudo-labeling, weak supervision, and synthetic data.
• Model Training and Experimentation: Experience with experiment tracking and scalable training tools such as PyTorch Lightning, DeepSpeed, or Ray.
• Model Optimization and Deployment: Familiarity with ONNX Runtime, TensorRT, CUDA, and C++ performance optimization for efficient inference on edge hardware.
• Mathematical Foundations: Strong understanding of probability, linear algebra, statistics, optimization, and deep learning fundamentals, with the ability to translate research into production-ready implementations.

Good to Have:

• Experience in 3D computer vision, point-cloud processing, depth estimation, or multi-view geometry.
• Familiarity with Open3D, ROS 2, Nav2, or MoveIt 2.
• Experience generating synthetic visual data using NVIDIA Isaac Sim or similar simulation platforms.
• Experience deploying real-time computer vision models on NVIDIA Jetson or other resource-constrained edge devices.
• Publications in computer vision, machine learning, or robotics conferences such as CVPR, ICCV, ECCV, NeurIPS, or ICLR.