AI Research Engineer, Computer Vision & Multimodal AI

About the job PREDICTION is looking for an exceptional AI Research Engineer to help build the core intelligence behind the platform. This is a highly technical, hands‑on role for someone who wants to develop real multimodal AI systems and solve difficult problems across computer vision, generative AI, document understanding, and structured technical design. The role is ideal for a graduate student, PhD student, recent graduate, or exceptional senior undergraduate with meaningful hands‑on machine learning experience. Responsibilities Build multimodal AI systems that understand fashion sketches, garment images, reference photos, written instructions, and technical documents. Develop sketch‑to‑design, image‑to‑design, and design‑to‑tech‑pack pipelines. Build models that identify garment attributes, including silhouette, fabric, color, patterns, stitching, seams, closures, pockets, collars, sleeves, trims, labels, and construction details. Generate structured tech‑pack outputs, including bills of materials, measurement tables, construction notes, colorways, tolerances, fabrics, trims, and manufacturing instructions. Develop computer vision pipelines for garment segmentation, component detection, visual grounding, similarity matching, and attribute recognition. Experiment with vision‑language models for visual reasoning, design interpretation, structured extraction, and multimodal generation. Fine‑tune and evaluate foundation models for specialized design and technical‑document workflows. Convert unstructured images, sketches, and written instructions into validated JSON, tables, specifications, and technical documents. Develop evaluation systems for accuracy, completeness, consistency, visual similarity, and hallucination reduction. Build proprietary datasets, annotation workflows, synthetic‑data pipelines, and model‑evaluation benchmarks. Integrate models into production APIs and customer‑facing product workflows. Optimize inference performance, GPU usage, latency, memory, and throughput. Help define the long‑term AI architecture, model strategy, and technical roadmap. Qualifications Strong background in computer vision, deep learning, generative AI, machine learning, robotics, or multimodal systems. Currently pursuing or recently completed a Bachelor’s, Master’s, or PhD in Computer Science, Computer Engineering, Electrical Engineering, Applied Mathematics, Robotics, or a related field. Exceptional recent graduates and senior undergraduate students are encouraged to apply. Experience in machine learning, computer vision, or applied AI with at least two years of hands‑on experience. Experience with image segmentation, object detection, image classification, visual similarity, feature extraction, image generation, or document understanding. Familiarity with vision‑language models, multimodal foundation models, large language models, or generative image models. Strong Python skills and experience with PyTorch or similar machine learning framework. Comfortable training, fine‑tuning, evaluating, and deploying models in GPU‑based environments. Ability to read current research papers and implement promising approaches. Ability to build complete systems rather than isolated notebooks. Comfortable working in a fast‑moving, early‑stage environment with significant ownership. Must be based in the New York City area or willing to relocate. Technical Skills Machine Learning Frameworks: PyTorch, TensorFlow, Hugging Face Transformers, Lightning, JAX. Computer Vision: OpenCV, SAM, SAM 2, Grounding DINO, Detectron2, YOLO, DINOv2, CLIP, SigLIP, image segmentation, visual embeddings, and similarity search. Multimodal AI: Qwen‑VL, LLaVA, Florence, Gemini, GPT vision models, or comparable vision‑language architectures. Generative AI: Diffusion models, image‑to‑image generation, controlled generation, inpainting, image editing, visual conditioning, and structured generation. Model Development: Supervised fine‑tuning, LoRA, embeddings, retrieval‑augmented generation, prompt engineering, confidence scoring, model routing, and human‑feedback workflows. Technical Document Understanding: OCR, PDF parsing, table extraction, diagram understanding, schema‑based extraction, structured JSON generation, and validation pipelines. Data and Evaluation: Dataset creation, annotation tools, synthetic‑data generation, model benchmarking, error analysis, data‑quality control, and human evaluation. Infrastructure and Deployment: Linux, Docker, NVIDIA GPUs, CUDA, cloud GPU environments, model serving, inference optimization, production APIs, and monitoring. Programming Languages: Python required; C++, JavaScript, or TypeScript is a plus. Additional Advantages Experience building an AI product from early prototype through production deployment. Experience with fashion technology, apparel design, garment construction, technical design, CAD systems, manufacturing, or tech packs. Experience working with sketches, technical drawings, measurement tables, diagrams, or manufacturing documents. Research publications in computer vision, multimodal AI, generative AI, or related fields. Open-source contributions or technically impressive independent projects. Experience at a leading AI laboratory, technology company, university research group, or early‑stage startup. Location This is a hybrid position based in New York City. Candidates must be available to work regularly with the team in person at our Times Square office. This is not a fully remote position. #J-18808-Ljbffr PREDICTION