Niantic Spatial, Inc. fizikaviy dunyoning mashinalar o‘qiy oladigan tasvirlarini yaratish uchun geofazoviy sun’iy intellekt ishlab chiqadi. Kompaniya platformasi yuqori aniqlikdagi, geografik jihatdan bog‘langan 3D raqamli egizaklarni yaratish, GPS ishonchsiz bo‘lgan holatlarda vizual joylashuvni aniqlashni qo‘llab-quvvatlash va xaritaga tushirilgan muhitlarga semantik ma’no qo‘shish uchun yerusti hamda havo sensorlari ma’lumotlarini birlashtiradi. Makonni tasvirga olish, qayta tiklash, joylashuvni aniqlash va anglashga mo‘ljallangan mahsulotlari orqali kompaniya robototexnika, davlat sektori va mudofaa, sanoat hamda yirik tadbirlar o‘tkaziladigan obyektlardagi foydalanish holatlari uchun SDK vositalari va korporativ xizmatlarni taqdim etadi.
Computer Vision AI Engineer
Train computer vision localization models for Niantic Spatial’s physical AI and geospatial platform. Build data pipelines, evaluation benchmarks, and production inference systems.
Tavsif
- Develop, train, and assess visual localization models using PyTorch.
- Refine model architectures, loss functions, and training loops to improve accuracy on production imagery.
- Help build high-throughput pipelines to ingest and preprocess images, pose data, and 3D point clouds.
- Run model training on GPU systems using a corpus of 30 billion images.
- Build and maintain benchmarks that measure localization pose accuracy, recall, and production latency.
- Prepare models for deployment through export, inference optimization, and monitoring.
- Partner with Research and the Real-World Test Lab to turn field failures into training data and benchmark improvements.
- Contribute throughout the pipeline, from data preparation and training to production services.
Talablar
- Experience training PyTorch deep learning models for computer vision through coursework, research, internships, or early industry roles.
- Foundational knowledge of at least one area: visual localization, learned feature matching, image retrieval, pose estimation, SLAM, or structure-from-motion.
- Understanding of 3D geometry, including camera models, epipolar geometry, and pose representations.
- Comfort working with large image or 3D datasets, including their preprocessing and curation.
- Familiarity with GPU-based training and the ability to assess training runs.
- Strong Python skills.
- Bachelor’s or master’s degree in computer science, computer vision, robotics, or a related field, or equivalent practical experience.
- Ability to work from the San Francisco office at least three days per week.
Imtiyozlar
- Annual bonus
- Equity
- Medical coverage
- Dental coverage
- Vision coverage
- 401(k)
- Comprehensive benefits package