About The Position

Centific is hiring an AI Research Engineer focused on Vision AI, vision-language models, and physical AI. The role is based in Seattle, Washington, or may be remote, and focuses on translating research into production systems that perceive, reason, and act in real-world settings. The engineer will own experiments from paper through prototype to deployable platform modules, working across 2D and 3D perception, egocentric and 360-degree understanding, multimodal reasoning, and embodied intelligence. The position may involve work with the Computer Vision team on 3D reconstruction, scene understanding, and visual AI, or with the Physical AI Robotics team on simulation, robotics, and AI. The role uses research and engineering approaches to build perception models and VLMs, develop perception-in-the-loop policies, curate data and evaluation protocols, and package systems for reliable deployment. Work may also include agentic pipelines combining perception, reasoning, simulation, and code generation. Centific describes itself as a frontier AI data foundry serving enterprise clients and major technology companies. Its teams develop contextual datasets, fine-tuned industry-specific LLMs, and RAG pipelines. The company’s mission is to bridge AI creators and industry leaders and help organizations deploy generative AI at scale.

Requirements

  • Master’s or Ph.D. in computer science, electrical engineering, robotics, or a related field.
  • Active publication record in computer vision, machine learning, or robotics venues such as CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR, CoRL, or RSS.
  • Strong Python and PyTorch or JAX skills.
  • Comfort with CUDA profiling and mixed-precision training.
  • Demonstrated research in computer vision and at least one of vision-language models, embodied/physical AI, or 3D perception.
  • Ability to carry research from paper to code, ablation, and results with rigorous experiment tracking.

Nice To Haves

  • Experience with video models such as TimeSFormer, MViT, or VideoMAE; diffusion or 3D Gaussian Splatting/NeRF pipelines; or SLAM/scene reconstruction.
  • Prior work on multimodal grounding, including referring expressions, spatial language, or affordances, or on temporal reasoning.
  • Familiarity with ROS2, DeepStream/TAO, or edge inference optimization using TensorRT or ONNX.
  • Experience with scalable training, Ray, distributed data loaders, or sharded checkpoints.
  • Strong software engineering practices, including testing, linting, profiling, containers, and reproducibility.
  • Public code artifacts such as GitHub projects, first-author publications, or substantial open-source impact.
  • Experience with NVIDIA Omniverse, Isaac Sim, GR00T, Holoscan SDK, Newton Physics, or NVIDIA NERD.
  • Experience with tools and frameworks including Open3D, PyTorch3D, MuJoCo, COLMAP/SLAM, NeRF/3DGS, FastAPI, Kubernetes, Docker, Triton/TensorRT, Weights & Biases, and LangGraph-like orchestration.

Responsibilities

  • Build and fine-tune models for detection, tracking, 2D/3D segmentation, pose and activity recognition, and scene understanding, including 360-degree and multi-view settings.
  • Train and evaluate vision-language models for grounding, dense captioning, temporal question answering, and tool use; design retrieval-augmented and agentic loops for perception-action tasks.
  • Prototype perception-in-the-loop policies using simulation and real data, and integrate them with planners and task graphs for manipulation, navigation, or safety workflows.
  • Curate datasets, develop evaluation protocols and KPIs, and run ablation studies.
  • Package research into reliable services using systems such as Kubernetes, Docker, Ray, and FastAPI, with profiling, telemetry, and CI for reproducible science.
  • Orchestrate multi-agent workflows that combine perception, reasoning, simulation, and code generation for self-checking and self-correction.
  • Conduct research and engineering across 3D reconstruction, scene understanding, visual AI, and/or robotics simulation and training.
  • Produce publishable or open-source work, subject to company approval, or production-ready modules that measurably improve product KPIs.
  • Maintain clean, reproducible code, document ablations, provide evaluation reports, and create capability demonstrations.

Benefits

  • Mentorship and close collaboration with the Principal Architect and senior engineers/researchers.
  • Opportunity for research to be deployed in company MVPs and products.
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service