Research Scientist- SLAM & VIO

The role

This person will architect and train proprietary foundation models from scratch. They will build 3D reconstruction architectures and develop robust optical flow models for egocentric vision using proprietary ground-truth data.

What you will do

• Design, implement, and train feed-forward network and per-scene differential optimization architectures for 3D geometry extraction.
• Scale multi-view ML architectures across multi-GPU clusters.
• Develop novel loss functions and attention mechanisms.
• Build and train custom optical flow architectures for egocentric motion modeling.
• Use flow models to segment dynamic actors and track objects.
• Rapidly prototype and deploy new models for tracking, segmentation, and multi-modal sensor fusion.
• Resolve algorithmic bottlenecks in the data engine.
• Connect foundational model outputs into dense reconstruction pipelines like 3D Gaussian Splatting and NeRFs.

Required qualifications

• Deep expertise in Deep Learning, 3D Computer Vision, and Temporal/Video Modeling.
• Proven experience training large-scale vision models from scratch.
• Strong theoretical and practical understanding of modern feed-forward 3D networks and dense motion estimation.
• Mastery of PyTorch and deep learning scaling frameworks.
• Experience handling and curating massive, multi-terabyte image and video datasets for training.
• Comfortable operating in a fast-paced environment where priorities can shift rapidly.

Nice to have

• First-author publications in top-tier venues (CVPR, ICCV, ECCV, NeurIPS) focusing on 3D deep learning, optical flow, video generation, or spatial transformers.
• Specific experience working with egocentric video datasets (e.g., Ego4D, Ego-Exo4D).
• Experience writing custom CUDA kernels to accelerate 3D operations, ray marching, or correlation volume computations.

About the company

Mecka AI is building the data infrastructure layer for robotics and embodied AI. They design and operate global systems for data capture, data labeling, and hardware-enabled workflows for AI labs and robotics companies.