Full-Time
Research Scientist
RL & Simulation
Mecka AI
51-200 employees
Labeled human movement data for robotics
Compensation Overview
$200k - $250k/yr
New York, NY, USA
In Person
Master's, PhD
 AI & Machine Learning (1) |
|---|
Get referred to Mecka AI
See people who can refer or advise you
- A Master of Science or Doctor of Philosophy degree, or equivalent research experience, in robotics, machine learning, or a related field.
- Strong hands-on experience with robot simulation and policy learning.
- Proficiency in Python and solid engineering discipline, including reproducible experiments, clean code, and debugging.
- Comfort working end-to-end across environments, data, training, and evaluation.
- Own the reinforcement learning and simulation engine that turns large-scale human demonstrations into scalable robot learning signals.
- Build and maintain simulation environments for robotics learning using platforms such as Isaac Sim, Isaac Gym, MuJoCo, Genesis, Habitat, or ManiSkill.
- Decide which environments and assets to build first to maximize learning velocity.
- Convert human demonstrations into robot-executable trajectories.
- Explore inverse-kinematics-based, optimization-based, and learning-based retargeting approaches.
- Train policies from demonstrations using imitation learning and reinforcement learning, including behavior cloning, DAgger-style aggregation, and offline reinforcement learning.
- Use proximal policy optimization, soft actor-critic, or similar methods when online fine-tuning is required.
- Define evaluation using success metrics, stress tests, generalization, and regression tracking.
- Drive sim-to-real transfer through domain randomization, system identification, contact modeling, and failure-mode analysis.
- Use real data to identify relevant domain gaps.
- Experience with manipulation, dexterous hands, or locomotion.
- Experience with retargeting, inverse kinematics, trajectory optimization, or differentiable simulation.
- Deep intuition for what makes sim-to-real transfer succeed or fail.
Mecka AI supplies large-scale, labeled human movement data to train AI and robotics systems. It records and analyzes videos of people performing targeted tasks, then delivers anonymized, richly annotated multimodal datasets that are ready for machine-learning models. The company coordinates a global network to collect data and provides end-to-end services from data collection to annotation and dataset curation, including a gig-economy model for participation. Its goal is to give AI developers and robotics teams fast access to high-quality movement data at scale to speed up autonomous-system development.
Company Size
51-200
Company Stage
Series A
Total Funding
$60M
Headquarters
Barbados
Founded
2025
Get referred to Mecka AI
See people who can refer or advise you
Simplify's Take
What believers are saying
- Framework Ventures led Mecka’s $60 million raise in June 2026, validating demand.
- Fortune reported already-signed contracts and a projected $100 million annual run rate.
- Mecka opened a New York office in 2026 and grew to 45 employees.
What critics are saying
- OpenAI, Google DeepMind, and robotics labs internalize data pipelines, compressing Mecka’s pricing power.
- Privacy regulators scrutinize filmed workers and sensor data; one EU enforcement action disrupts sourcing.
- Humanoid-robotics demand stays speculative; customer cancellations would kill Mecka’s projected $100 million run rate.
What makes Mecka AI unique
- Mecka AI sells bespoke human-motion datasets, not generic labeling, for physical-AI training.
- Its 12-country contributor network captures iPhones, body sensors, and custom-camera video at scale.
- Docula’s 2026 acquisition adds motion-data processing infrastructure for Mecka’s deployment layer.
Help us improve and share your feedback! Did you find this helpful?
Growth & Insights and Company News
6 month growth
↑ 26%1 year growth
↑ 26%2 year growth
↑ 44%Toronto- and New York City-based Mecka AI has acquired Vancouver startup Docula to help process massive volumes of motion data for training robots. The deal, completed early this year, brought all three members of bootstrapped Docula's team to Mecka. Docula had built an AI data-processing engine for medical billing that could handle large data volumes. Mecka aims to build "the data and deployment layer" for physical AI by collecting three-dimensional movement data from hundreds of thousands of contributors across 12 countries using iPhones and custom cameras. The US robotics market reached $11.4 billion in 2026, up nearly 30 percent year-over-year. Mecka recently opened a New York office to expand its US customer base, though over half its 45-person team remains Canadian.
Framework Ventures has led a $60 million funding round for Mecka AI, a company specialising in physical AI. The venture capital firm announced the investment whilst reaffirming its confidence in digital assets, noting its substantial holdings in Hyperliquid, Plasma and Sky. No additional details about Mecka AI's technology, valuation or other participating investors were disclosed. The funding marks Framework Ventures' continued focus on AI-related investments within the digital asset space.
Mecka AI announces $60 million USD to power physical AI. Canadian-led startup aims to build "the data and deployment layer" for robotics. Mecka AI, which aims to train robots with human data sourced from body sensors and iPhones, revealed Monday that it has raised $60 million USD in funding. The news: The Toronto and New York City-based startup wants to build "the data and deployment layer" for physical AI. Mecka secured this sum across two previously unannounced SAFE financings: an initial $25-million Series A round in November and $35-million Series A extension that closed in recent weeks. San Francisco's Framework Ventures, which typically focuses on crypto, led with support from SV Angel and ex-Google DeepMind researcher Ted Xiao, now a founding member of Jeff Bezos' AI startup, Project Prometheus, among others. From the source: "Robotic deployment use cases are messy in the real world and not all tasks make sense today," Mecka co-founder and CEO Josh Gao argued in a news release, adding that the startup brings the "last-mile" data processing, models, and other capabilities necessary to make robots work in the field. Following the thread: Mecka and others are betting that video captured from a first-person perspective will help robots not learn faster and scale more efficiently. The startup currently collects this data from home settings, culinary work, chemistry labs, task platforms, metal fabrication, and leather shops. Its internal "video understanding lab" turns this raw footage into training-ready data and refines its computer vision models. Since forming last year, the fast-growing firm has raised $68 million, begun catering to multiple frontier robotics labs and Big Tech firms, and become - Mecka AI claims - one of the world's largest providers of physical AI training data. Though it is domiciled in the US, three of its four co-founders are Canadian and 40 of its 45 staff are based in Toronto. It plans to use this money to help more companies deploy robots. Final thought: AI claimed nearly half of all venture capital dollars invested in Canada last year, and global interest in robotics startups, humanoid makers, and companies developing physical AI that integrates into physical systems has also been heating up. Other recent Canadian beneficiaries of this trend include Vancouver's A&K Robotics and Montréal-based Mecademic. The Canadian Robotics Council has clocked "a rapid increase" in the amount of robotics startups being launched, both at home and abroad. Roborock Discover the next generation of cleaning innovation. Roborock redefines home robotics with AI-powered systems that sense, adapt, and operate seamlessly in real homes.
Mecka AI, a New York City based robotics data and infrastructure company powering the next generation of physical AI, raised $60m in funding
Shareholders in private companies are holding their breath as SpaceX, OpenAI, and Anthropic look to make big splashes in the public markets.