The short version, and the long one

I am Apoorve, an AI and security engineer in Santa Clara. At Cohesity I work on identity threat detection and response: spotting attacks against the identity systems enterprises run on, Entra ID, Okta, Ping and Active Directory, and working out how far an attacker can get from wherever they got in.

I got here by a fairly long road. I studied computer engineering at Thapar Institute in Patiala and started out in Bengaluru, first at Arista Networks on network protocols, then at Amdocs building telecom microservices for Telstra, then at VMware on a vulnerability management platform that tracked over a hundred million user entries. That last job was where I made my first open-source contribution, to StackStorm, and where I realized I liked infrastructure that has to be right at scale.

In 2022 I moved to the United States for a master's in computer science at UNC Charlotte, with a focus on data science, and spent a year of it as a research engineer building bioinformatics tooling for federally funded biomedical work. After graduating I went back to VMware, this time through HCL America, on VMware Cloud on AWS, then AWS License Manager, where I built the compliance and asset-governance infrastructure that enterprise and federal customers depend on: License Asset Groups, the Resource Sweeper, and an anomaly detector for metering.

Outside work I research the half of AI supply chain security that nobody had written down: what an operator actually does once a model or dataset turns out to be compromised. That became a preprint in September 2026, a reproducible implementation, and an essay. The thread running through all of it, work and research alike, is the same question: given a graph of things that trust each other, what can a compromise reach, and what can you still fix?

If any of that sounds like something you are working on, email me. I read everything.