Anirban Kanjilal
Frontier compute, made secure, cost-governed, and production-real.
I take AI and quantum out of the demo and into regulated, always-on production. Twenty years building real-time payments taught me the hard part is rarely the model — it's the security, the cost, and the SLA.
Prompt injection shown, not told — vulnerable vs. hardened variants of real LLM risks, each mapped to OWASP LLM Top 10, MITRE ATLAS, and the NIST AI RMF. Runs offline with zero API keys.
Runnable, cost-aware quantum algorithms on Amazon Braket. Free on the local simulator, one flag to real hardware, with a cost guard before every QPU call. Grover today; QAOA and VQE next.
An agent that reads FOCUS-format billing data and surfaces where AI and cloud spend is going — idle GPUs, inference spikes, cost-per-outcome — with the agent itself hardened against injection.