Research Scientist, SLAM & VIO

Mecka•New York, NY

3d•$200,000 - $250,000

About The Position

Mecka AI is building the data infrastructure layer for robotics and embodied AI. We partner with leading AI labs and robotics companies to deliver high-quality, real-world datasets used to train, evaluate, and deploy robotic systems — where model performance is dictated by data quality. We are hiring a Research Scientist (SLAM & Visual-Inertial Odometry) to build and validate state estimation systems that work in the real world, on messy sensors, under tight compute and reliability constraints. This role is research-heavy but production-minded: you will ship algorithms that survive scale, long runtimes, and operational edge cases.

Requirements

Strong experience in SLAM / VO / VIO (academia or industry), with evidence of shipped systems or publishable results.
Solid understanding of estimation: nonlinear least squares, factor graphs, filtering/smoothing, and uncertainty.
Proficiency in C++ (and comfort in Python for research and evaluation).
You have built systems that run for hours/days and degrade gracefully, not just “works on a benchmark.”
You understand real sensor failure modes: calibration drift, sync issues, rolling shutter, motion blur, low light.
Experience with modern tooling (e.g., GTSAM/Ceres), and strong intuition for optimization and numerics.

Nice To Haves

Experience with learned front-ends/back-ends (e.g., learned features, depth, relocalization, or hybrid classical+ML pipelines).
Experience building offline mapping / batch optimization pipelines for large datasets.
Familiarity with embedded/edge constraints and profiling/optimization.

Responsibilities

Develop robust monocular VO/VIO pipelines (feature-based and/or learned) with strong failure detection and recovery.
Address scale ambiguity with inertial fusion, motion priors, and consistency constraints.
Ensure online performance: low latency, bounded memory, and stable tracking across lighting, motion blur, rolling shutter, and dynamic objects.
Build offline reconstruction pipelines for long trajectories: global BA, loop closure at scale, and map optimization.
Produce high-quality trajectories and sparse/dense maps for downstream data products (labeling, QA, training signals).
Design evaluation tooling: drift decomposition, per-segment error, and systematic bias detection.
Implement stereo VO/VIO with accurate calibration handling (intrinsics/extrinsics, temporal sync) and robust matching.
Improve depth reliability across challenging scenes (low texture, repetitive patterns, specularities).
Optimize for stability and long-duration runs: track health metrics, relocalization, and graceful degradation.
Perform large-scale mapping and trajectory refinement using stereo constraints.
Conduct loop closure + global pose graph optimization with principled uncertainty handling.
Produce maps that are useful, not just pretty: consistent frames, repeatable landmarks, and clear quality scores.
Work on sensor modeling & calibration: rolling shutter, time offsets, IMU noise/scale factors, and temperature-driven drift.
Focus on robustness engineering: automatic resets, outlier handling, and “what broke?” diagnostics.
Design evaluation suites, curate failure cases, and define release gates.