Algorithm Engineer - REM
We're building a lightweight 2D vector map system for intelligent driving and the next-generation map reconstruction stack. We adopt learning-based algorithms to reconstruct structured road layers from mass vehicle driving data. Our team combines computer vision, topological / graph learning, and generative spatial modeling to build fully automated map production pipelines, with rapid iteration as our core value.
What will you job look like:
- Develop learning-based algorithms to reconstruct structured road vector data and next-generation map outputs using mass crowdsourced vehicle perception records and multi-modal sensor inputs.
- Model road geometry, semantic features, lane connections and global road topology through spatial reasoning, topological learning networks and graph networks.
- Combine deep learning and graph modeling with generative methods (e.g. diffusion, structured prediction) and 3D spatial reconstruction to tackle complex urban scene challenges.
- Write standardized, maintainable and testable production code with Python/C++, participate in code review and drive team technical iteration.
All you need is:
- Master or Ph.D. in Computer Science, Electronic Engineering, Robotics or related majors.
- 2+ years algorithm development experience in computer vision, topological / graph learning, generative AI, spatial modeling, trajectory mining.
- Comfortable with basic geometry and spatial data representation (coordinates, curves, connectivity).
- Experience with at least one of: topological learning networks, generative models (diffusion / flow matching), or 3D point-cloud / scene reconstruction.
- Solid programming and algorithm capabilities with Python or C/C++; proficient in at least one deep learning framework (PyTorch / TensorFlow preferred).
- Fluent oral and written communication in both Mandarin and English, excellent team player.
Nice-to-have
- In-depth understanding of CNN, GNN, Transformer, object detection, semantic segmentation and generative AI.
- Familiar with topological learning networks like MapTR, and related lane / road topology modeling methods.
- Experience with diffusion models, generative AI, or structured output generation for maps, layouts, graphs, or splines.