Mecka AI is building the data infrastructure layer for robotics and embodied AI.
We design, deploy, and operate real-world robotic systems to generate high-quality datasets used to train and evaluate next-generation robotics and embodied AI models. Our robots operate in unstructured, real-world environments, where reliability and execution matter more than clean simulations.
We’re hiring a Roboticist with a strong bias toward execution and deployment.
This is not a theoretical research role. This is a “make it work in the wild” role. You will take high-level models, perception pipelines, and navigation policies and deploy them onto physical robots. Your job is to close the gap between simulation and reality, ensuring robots can navigate, manipulate, and collect data autonomously in real environments.
You’ll spend time debugging in the lab and on-site, diagnosing failures, and building systems that keep running even when things go wrong.
Integrate Vision-Language-Action (VLA) models, perception pipelines, and control policies into a cohesive robotic stack
Write the glue code that translates model inference into physical motor actions
Ensure models run reliably and predictably on real hardware
Implement, tune, and maintain:
SLAM
Localization
Path planning
Obstacle avoidance
Ensure smooth and intelligent robot motion in dynamic, human environments
Build resilient, self-healing systems
Develop watchdogs, automated calibration routines, and error-handling logic
Enable robots to recover from common failures without human intervention
Debug robots in the lab and in real deployment environments
Diagnose issues such as:
Failed docking or navigation
Sensor drift or misalignment
Policy failures or hallucinations
Identify root causes across software, sensors, and hardware — and fix them
Strong software engineering skills in Python and C++
Experience writing production-grade code for robotic systems
Comfortable working in Linux environments (processes, networking, system configuration)
Experience with pub/sub and message-passing systems such as:
Tool-agnostic mindset — you understand the concepts, not just the framework
Systematic approach to debugging electro-mechanical systems
Ability to isolate failures across:
Software
Sensors
Networking
Hardware
Comfortable getting hands-on when things break
Work directly on real robots operating in the physical world
Own deployment, reliability, and execution — not just models
Solve hard, messy problems at the boundary of software and hardware
High ownership and autonomy in a fast-moving environment
See your work directly impact data quality and system performance
