A research and teaching lab working on how to keep increasingly capable AI systems safe, secure and under meaningful human oversight.
In the 1940s, a quiet group of mathematicians and engineers at Bletchley Park’s Station X took on an adversary more capable than any single person. Among them was Alan Turing, whose work went on to found computer science and artificial intelligence.
Station Zero carries that spirit to Mumbai. Zero is India’s gift to mathematics, and the slashed zero is the engineer’s mark. We study how to understand, test and secure AI systems before they outpace our ability to oversee them.
Keeping AI goals and values on course as systems grow more capable.
Attacking our own systems first: jailbreaks, prompt injection and adversarial testing.
Seeing inside the black box to understand why models do what they do.
Securing model weights, training data and the infrastructure around them.
Policy, standards and oversight, with a view from India and the Global South.
Sandboxing, monitoring and safe deployment of autonomous agents.
AI red teaming, Indic adversarial testing, a sealed air-gapped test range and regulatory assurance for banks, brokers, insurers and the AI vendors that serve them.
Controller-level testing, firmware analysis and readiness for India’s binding power-sector cyber rules, for OEMs, generators and the investors behind them.
A hands-on course where students attack, defend and audit real AI systems, and learn the security thinking that the next decade of engineering will demand.
Technology executive across AI, digital-asset security and autonomous systems: production ML at Tesla, threshold MPC custody at ZebPay, R&D at Keep Network and research at Georgia Tech Research Institute. CISM; MS Engineering and MBA, Georgia Tech.
[Short bio and research interests]
The lab’s first Hut members will be listed here once the cohort is selected.