The lab

An independent lab, run hands-on and backed by sharp advisors.

Nativerse is run by Srinathprasanna Shanmugam, with a small group of advisors across data science, AI research, backend and go-to-market. Everything below is shipped and public, from a live data product to a security audit maintainers acted on.

Srinathprasanna Shanmugam

Founder and lead researcher, London

Behavioural verification Live telemetry Systemic security Enterprise software & GTM

More than a decade inside enterprise software, from IBM and AT&T to fast-moving fintechs, banks and ISVs across Europe. Years spent watching how real systems break under scale, compliance and integration pressure. Nativerse points that experience at the security and verification of AI systems: equiv, the GAWK observatory, and the identity research. Nearly all of it is published, so every claim comes with a link you can check.

The work, by track.

Three tracks carry the lab, each anchored by something running today. A few sharper utilities sit below them. Each links to something you can open or read for yourself.

Track 1: Behavioural verification

equiv

Differential execution: run a changed function and its previous version on identical inputs and prove whether the behaviour held, with a signed receipt that re-runs byte-for-byte. Proven on AI-written code; the same method extends to comparing model behaviour.

Track 2: Live telemetry

GAWK

Verified public data on how the AI ecosystem moves, from model releases to spend across more than thirty sources. Every figure links to its origin, and a dropped source greys out and keeps its last value instead of guessing. Cited, not invented.

Track 3: Systemic security

Identity at scale

A scan of over two million repositories found 124 keying Google login on email instead of the stable OIDC sub claim. authdrift packages the check as six Semgrep rules; maintainers have merged fixes. The research direction is how this fails as agents inherit legacy auth.

Utility lab artifacts

Sharper tools

Small, deterministic tools built to solve isolated bottlenecks, open-sourced for the community: bharataddress (offline Indian address parsing), privacylint (App Store privacy checks), and codepulse (a CLAUDE.md audit).

People who keep the work honest.

A small group I lean on, from data science and AI research to backend and go-to-market.

SS

Srinivas Sampath

Development Manager at Cognizant.

LinkedIn
VV

Venkhat Vairavan

Senior Data Scientist at Nutanix. IIT Bombay.

LinkedIn
SR

Suresh Kumar Ravi

Backend Lead at Caterpillar.

LinkedIn
SN

Sabari Nathan

Senior AI Engineer at Couger, Tokyo. Published at CVPR, ECCV and ICCV.

LinkedIn
RP

Raj Prasanna Ponnuraj

AI go-to-market lead at Oracle. UT Dallas.

LinkedIn
VR

Vasanth K Rajendran

Tech Lead at KLA Tencor.

LinkedIn

If you work in one of these areas and something here overlaps with yours, get in touch.

Start a conversation