About Gerra
We originate the data
that doesn't exist yet.
Frontier models have read the open web. The next gains, in markets, in agents, and in the physical world, come from data it doesn't contain. Gerra builds access to that data: proprietary, real-world, and sourced from places no one else can reach.
Why we exist
The data wall is real.
The public internet has been scraped, and the marginal training corpus is largely exhausted. What is scarce now is proprietary signal: real trading behavior, real engineering and operational data from live companies, and real robot interaction with the physical world.
That is the gap we fill. We originate, license, and develop exactly that data, then deliver it backtest-ready for systematic funds and training- and eval-ready for AI labs. Not an aggregator. Not a scraper. A data-origination company.
How we got here
From robots to data.
Gerra started as a humanoid-robot operation. We built collection infrastructure, ran teleoperation, and captured multi-modal motion data at scale. The robots drew the attention; the data they produced was the actual asset.
In mid-2025 we made that the company, pivoting from deploying hardware to originating the data the physical world, and then the markets and the software world, produce. Today that spans three domains, each built from sources only we have.
What we do
Market Intelligence
Proprietary alternative-data signals for systematic funds, from social and sports to the compute markets behind AI.
Enterprise & Code Intelligence
Real workflow, codebase, and full-company data for training and evaluating models and agents, not synthetic benchmarks.
Physical AI
Multi-modal data from our own robot fleet: teleoperation, human demonstration, and sensor streams for embodied AI.
How we work
We originate, we don’t aggregate.
Anyone can resell a feed. We build the collection itself - robot fleets, exclusive platform licenses, consented company archives - so the data exists because we made it.
Exclusive access, not reseller terms.
Where we license, we hold exclusive, multi-year rights. If a buyer could get it elsewhere, we don’t bother sourcing it.
Research-grade rigor, in public.
We publish our methods and results openly, including the ones that fail - like a GPU-pricing signal that does not yet survive honest validation. We would rather report a null result than sell false alpha.
The work so far
Team
Ojas Shukla
Founder & CEO
Ojas has worked at the edge of data and markets since he was a teenager: trading equities at 16, publishing research at 17, and exiting his first company to SAP at 19. Before Gerra he built petabyte-scale encrypted data infrastructure and traded over $1B as a lead quantitative trader, running the largest automated market maker on Polymarket. He founded Gerra after operating one of the largest commercial humanoid fleets in the US, when it became clear the data, not the robots, was the asset. He still writes the research behind the datasets the company ships.
Aryan Mahajan
COO
Aryan runs Gerra's operations and commercial partnerships, turning originated data into live relationships with funds, labs, and data marketplaces. An AI architect by background, he has designed B2B AI systems for enterprises and brings a deep operator network across the data and finance worlds. He owns the path from dataset to deployment: sourcing demand, structuring licenses, and keeping delivery on time.
Backed by data-collection teams across India and the Philippines.
Work with us.
If you are training frontier models, running systematic strategies, or building embodied AI, we likely have data you cannot get anywhere else.