Shared evaluation
Honest, reproducible benchmarks make the whole field better. We’d rather compete on results than on secret test sets.
HomeCompanyOpen source
CompanyExplore the tools we can share: public repositories, permissive licences and practical ways to contribute.
01Projects
Every public repository in the Newron org, with its licence and last activity. Nothing here is listed as maintained that we are not touching.
Electron-based crawler for automating repetitive collection and scraping tasks.
Framework for running data collection at scale.
SDK for the data-centric ML platform Newron started on — build, manage and deploy models through data-driven development.
02Why we do it
Our edge is in deployment and custom models — not in hoarding the basics.
Honest, reproducible benchmarks make the whole field better. We’d rather compete on results than on secret test sets.
Tooling for Indic OCR and speech is scarce, and most of ours is still customer-specific. Where we can separate it from customer data, releasing it lowers the barrier for everyone building here.
Open schemas for statements and documents would let institutions move data without being trapped by any one vendor. We’d like to publish ours; it isn’t public yet.
Customer-trained models, customer data and anything covered by a deployment agreement stay closed. We’d rather say that plainly than imply a bigger public footprint than we have.
03Contributing
Issues and pull requests are welcome on any public repository.
Start with an issue
Open one describing the bug or proposal before sending a large change, so we can align on approach.
Check the repository README
Setup, style and test expectations live there; open an issue if anything is missing or out of date.
Add to the benchmarks
New evaluation cases — especially for Indian languages and document types — are some of the most valuable contributions.
Sign your commits
We use the Developer Certificate of Origin; a signed-off commit confirms you can contribute the code.
04Questions
Per project, and it is listed next to each one above: cortex and n00bs are MIT, newron-sdk is Apache-2.0. The LICENSE file in each repository is the authority.
Yes. Issues and pull requests are welcome; start with an issue for anything substantial so we can discuss the approach.
The shared tooling, schemas and benchmarks are open. Our custom, customer-trained models are not — those are licensed to the customer.
Please disclose responsibly via our security contact rather than a public issue. See the Security page for details.
Build with us
Open an issue or send a pull request on anything public. If you’re building regulated AI in India, we’d love to compare notes.