Senior Applied Research Engineer, 3D Computer Vision

Das ist der Job

Optimize 3D systems for speed, accuracy, and efficiency at production scale.

Darum lohnt es sich

Drive integration of modern reality capture approaches into our production stack – making the calls on what’s ready to ship and what isn’t. Contribute to the team through code review, pairing, and design feedback. Author agent skills or tooling that other engineers use; contribute to the team's shared skills library.

Conduct rigorous evaluations of new AI tools and bring useful patterns to the team. Track the AI tooling landscape and bring useful patterns to the team. Hands‑on experience with modern reality capture – specifically, gaussian splatting and/or foundation 3D reconstruction models (MASt3R, VGGT, or comparable).

Fluency in modern ML frameworks (PyTorch, TensorFlow, or equivalent) and modern training stacks. Responsibilities Design, train, evaluate, and ship 3D reconstruction systems across our pipeline – gaussian splats, foundation 3D models, SfM, MVS, monocular depth, mesh reconstruction.

Own the hardest technical investigations in your area, from initial triage through production rollout and long‑tail support. Use the right tool for the problem – classical 3D computer vision when it wins, learned approaches when they win. Stay current with 3D vision research and evaluate promising techniques against our workflows.

Hold a high technical bar for your own work – high‑quality designs, well‑tested code, production‑ready ship habits. Codify debugging and investigation playbooks into reusable skills. Use AI tools daily across the SDLC, with judgment on where they help and where they don’t. Review agent‑generated code with the same rigor as human‑written code.

Qualifications Bachelor’s, Master’s, or PhD in Computer Science, Engineering, or a related field, with 5+ years of professional experience in 3D Computer Vision or 3D Machine Learning, or a PhD in Computer Vision, Machine Learning, or a closely related discipline.

At least one of these from real hands‑on work (production, startup, or other industry experience), not just paper familiarity. Demonstrable track record of building and shipping 3D reconstruction systems – in production, at a startup, or in another applied industry setting (not just research or prototype work).

Deep experience across the broader 3D perception toolkit – at least three of: feature detection and matching, SfM, MVS, monocular depth estimation, mesh reconstruction, SLAM. Published research or open‑source contributions in any of these areas is a plus.

Comfort across the classical‑vs‑learned spectrum – you reach for the right tool, not the trendiest one. A real track record with agentic development. Strong C++ proficiency – much of our photogrammetry pipeline runs in C++ and you’ll be working in it daily.

Strong ability to timebox experiments, iterate effectively, and triage routes to success when the path isn’t obvious. Ability to work as an effective remote engineer with AM standup overlap with PST. Strong written and verbal communication; you can take a technically dense investigation and make it land with PMs, leadership, and other engineers.

Open‑source agent skills, plugins, or prompts that others use. Experience running and monitoring many concurrent ML experiments in cloud environments. Comfort with cloud training and inference (AWS, GCP, Azure) and containerization (Docker, Kubernetes). Experience with geospatial systems or aerial imagery pipelines. #J-18808-Ljbffr