# Mecka.AI > Mecka.AI is a physical AI data infrastructure and robotics deployment company that collects and analyzes real-world human activity to train, evaluate, and support the deployment of robots. - Canonical URL: https://iq.wiki/wiki/meckaai - Categories: Organizations - Tags: AIInfrastructure - Created: 2026-10-21T15:02:48.351Z - Last updated: 2026-10-21T15:56:04.466Z - Source: IQ.wiki — the world's largest blockchain and crypto encyclopedia (https://iq.wiki) --- **Mecka.AI** is a physical artificial intelligence (AI) data and deployment company co-founded by [Josh Gao](https://iq.wiki/wiki/josh-gao), [Jason Chong](https://iq.wiki/wiki/jason-chong), Duy Nguyen, and Mogen Cheng. It builds real-world data infrastructure and supports enterprise robot deployment; it does not manufacture robots. Instead, it combines data infrastructure, hardware, models, and integration services to support robotics teams and enterprises.[\[8\]](#cite-id-cybjwew6er)​[\[1\]](#cite-id-83d6wez79a)​[\[4\]](#cite-id-8tj5yxz3rq)​ ## Overview Mecka describes itself as a data and deployment layer for physical AI. It collects and structures real-world human activity for robotics and AI teams, and supports evaluation and enterprise deployment of robots through data, hardware, models, and integration services.[[1]](#cite-id-83d6wez79a) ## History Mecka was established in 2024. Dealroom describes the company as having early roots in [cryptocurrency](https://iq.wiki/wiki/cryptocurrency) and says it explored using [tokenomics](https://iq.wiki/wiki/tokenomics) to decentralize data collection before focusing on robotics data.[\[2\]](#cite-id-tdzotn103u)​[\[3\]](#cite-id-xwwbfu3h6d)​[\[5\]](#cite-id-60u14qv1eg)​ In August 2025, Mecka announced an $8 million seed round led by [Neo](https://iq.wiki/wiki/neo), with SV Angel and Offline VC among the participants.[\[5\]](#cite-id-60u14qv1eg)​ On June 1, 2026, Mecka reported raising $60 million in a Series A led by [Framework Ventures](https://iq.wiki/wiki/framework-ventures).[\[8\]](#cite-id-cybjwew6er)​ In June 2026, Mecka acquired Docula. The company said Docula’s Rohan Parmar, Saurab Dhir, and Mikhail Sinitcyn would join Mecka; coverage described the acquisition as part of its work to build a data layer for robotics and process motion data.[\[8\]](#cite-id-cybjwew6er)​ On October 7, 2026, Mecka announced a $60 million Series B led by Sequoia Capital. Investors included NVIDIA, M12, Microsoft’s venture fund, Qualcomm Ventures, and Samsung, while existing investors Kindred, [Framework Ventures](https://iq.wiki/wiki/framework-ventures), and [Neo](https://iq.wiki/wiki/neo) continued their support. Angel investors included DoorDash founder and chief executive Tony Xu, former ServiceNow and Snowflake chief executive Frank Slootman, and former Tesla Optimus head Milan Kovac. Mecka said the funding would expand its data infrastructure and internal research lab and support commercial robot deployments. The company’s X account said the round brought its total funding to more than $120 million.[\[4\]](#cite-id-8tj5yxz3rq)​[\[6\]](#cite-id-qmjq2h9z7t)​[\[2\]](#cite-id-tdzotn103u)​[\[7\]](#cite-id-pgf3dsfmrs)​ ## Data Infrastructure and Collection Mecka collects human demonstrations in laboratories and real-world environments, including homes and commercial spaces. Participants record themselves performing everyday tasks—such as making coffee or fixing cars—using smartphones and, in some collection programs, body sensors or other capture hardware.[[2]](#cite-id-tdzotn103u)[[5]](#cite-id-60u14qv1eg)[[4]](#cite-id-8tj5yxz3rq) The company processes raw recordings into datasets for robotics and AI teams. Its workflows include annotation, contextualization, quality checks, motion understanding, camera and sensor calibration, and computer-vision analysis. Mecka’s web tools let teams browse, visualize, query, and collaborate on datasets, while inference APIs support model testing. Its data-collection product is also available on iOS.[[1]](#cite-id-83d6wez79a)[[3]](#cite-id-xwwbfu3h6d)[[5]](#cite-id-60u14qv1eg) Mecka describes its technology stack as combining capture hardware, reconstruction models, and quality-control infrastructure. It designs and manufactures multi-sensor hardware, selecting and synchronizing sensors according to the requirements of downstream models. Its research team develops computer-vision and multimodal models that turn recordings into structured outputs, including motion tracking, three-dimensional reconstruction, and sensor alignment. The company reports sub-centimeter hand-pose accuracy on in-the-wild data.[[4]](#cite-id-8tj5yxz3rq) Structured data can include recorded actions and associated context, but its contents vary by dataset and collection program. For EgoVerse specifically, the data includes egocentric video, 3D hand-keypoint trajectories, camera and head poses, task and scene/object metadata, and fine-grained task-language descriptions. EgoDB standardizes and indexes the data so filtered subsets can be used to train and evaluate robot policies.[[9]](#cite-id-yk88lh4com) Mecka’s privacy policy, effective March 20, 2026, covers information submitted through its services, including dataset contributions. It says the company does not sell personal information, describes TLS encryption and access controls among its security measures, and says personal information is retained as needed and securely deleted or anonymized when no longer needed. Depending on the circumstances, rights may include access, correction, deletion, restriction of processing, and withdrawal of consent where processing is based on consent.[[10]](#cite-id-zhkoh9e2f2) ## Research EgoVerse is Mecka’s large-scale dataset of human interaction, built from first-person recordings of activities across tasks, spaces, and environments. It is intended to help researchers study how people move, manipulate objects, and complete tasks, and to support robot learning from egocentric data.[[1]](#cite-id-83d6wez79a) Mecka tested the use of human demonstrations for robot learning through the EgoVerse human-to-robot transfer study, conducted with researchers at Georgia Tech, Stanford, the University of California, San Diego, ETH Zürich, the Massachusetts Institute of Technology, and Meta. The company says the study was replicated across laboratories, tasks, and robots.[[4]](#cite-id-8tj5yxz3rq) The study tested co-training robot manipulation policies on human egocentric demonstrations and robot demonstrations, with experiments across multiple labs, tasks, and robot embodiments. The paper reports that EgoVerse-A co-training improved in-domain and out-of-domain performance by up to 30% in the tested settings. It also reports that effective scaling of diverse human data depended on including aligned human–robot data.[[9]](#cite-id-yk88lh4com) ## Evaluation and Deployment Mecka offers model evaluation and helps teams test robotic systems against real workflows, environments, and constraints before field deployment. Its services are aimed at robotics laboratories, frontier-model teams, and enterprises working with embodied AI. Teams can use existing datasets, commission task-specific collection, run evaluations, or develop data programs across multiple environments.[[1]](#cite-id-83d6wez79a) The company also integrates robots for enterprise customers that lack in-house robotics teams. Its described deployment process combines hardware, on-site data capture, post-training, integration, and ongoing operations. Mecka says this approach allows robots to be post-trained on data from the deployment site and improved as they operate, rather than being delivered as fixed systems.[[4]](#cite-id-8tj5yxz3rq) Mecka reported that it exceeded $100 million in run-rate revenue in June 2026, after a few months of operations. In its June 2026 announcement, the company projected a $300 million run rate by the end of 2026.[[4]](#cite-id-8tj5yxz3rq) The company says it supplies several robotics research laboratories and multiple companies in the “Mag 7.” Dealroom identifies humanoid-robotics company 1X Technologies as an early partner.[[4]](#cite-id-8tj5yxz3rq)[[5]](#cite-id-60u14qv1eg) ## Leadership Josh Gao is Mecka’s co-founder and chief executive officer, and Jason Chong is its co-founder and chief technology officer. Duy Nguyen and Mogen Cheng are also identified as co-founders, and Eric Lok as a founding member.[[8]](#cite-id-cybjwew6er)