Quidient is a deep tech AI company pioneering advancements in Generalized Scene Reconstruction. They are seeking a Senior Offline Mapping Engineer to enhance the accuracy and robustness of their offline mapping and 3D reconstruction systems, utilizing both classical methods and deep learning techniques.
Responsibilities:
- Develop and advance our offline mapping and reconstruction pipeline, driving accuracy and robustness across the hardest capture scenarios — textureless walls, highly reflective surfaces, featureless geometry, and large-scale scenes
- Reduce pose estimation failures and increase geometric accuracy in environments where classical methods degrade, using a combination of improved estimation and learned components
- Design and integrate deep learning methods (learned feature matching, monocular depth priors, learned outlier rejection) alongside classical SfM and MVS components to close the last 10% of reconstruction quality
- Improve calibration pipelines, bundle adjustment robustness, and dense reconstruction fidelity in offline processing contexts where throughput matters but hard real-time does not
- Build and maintain evaluation methodology grounded in real-world captures — covering feature-rich, textureless, and reflective environments — not synthetic benchmarks alone
- Stay current with the deep learning and 3D vision literature, applying good judgment about which methods are production-viable and which are benchmark artifacts
- Collaborate closely with the SLAM and real-time mapping team to share components and ensure offline improvements feed back into the broader reconstruction platform