Jay Shah
A personal blog. Notes on engineering, work, and whatever I'm thinking through as I go.
Selected writing
Index →Projects
All →Argus
An internal AI knowledge assistant that turns a company's scattered knowledge across GitHub, docs, the wiki, Jira, and Slack into synthesized, source-cited answers in seconds, and gets sharper over time through a human-reviewed memory.
Vera
An agentic AI platform for privacy and compliance teams. Vera doesn't just answer questions; it takes real action, reading and writing across a customer's privacy program through governed tools. The guiding principle was agentic power with safety: customers stay in full control, and a human-in-the-loop model approves every consequential action. I led the team that built it as engineering manager while staying hands-on in the architecture and code, designing core parts of the agent and building many of the MCP tools it acts through.