$
Engineer. Technical leader. Builder.
I build security systems, infrastructure, and tools for AI agents. My experience spans backend development, architecture, production operations, and leading engineering teams at F5 and Shape Security. I stay close to the code and take responsibility for how systems work once they ship.
Production engineering
F5 / Shape Security · Web application bot detection
At F5, I owned the web application bot detection product end to end, leading architecture, development, testing, and deployment across AWS and GCP. Earlier, as a Staff Software Engineer at Shape Security, I designed and implemented backend features for the Anti-Bot Mitigation Engine and built internal engineering tools.
As the scope grew, I led a global organization of 15+ people, including three managers. I worked directly with customers to simplify configuration and led improvements to deployment, testing, monitoring, and infrastructure cost.
About F5 Bot Defense (opens in new tab)
Career
F5 · Senior Manager, Engineering · Apr 2021–Sep 2026
Product ownership, architecture, delivery, and engineering organization leadership.
Shape Security · Manager, Engineering · Apr 2020–Apr 2021; Staff Software Engineer · Nov 2018–Apr 2020
Backend bot mitigation, cloud operations, APIs, and engineering tooling.
Apple · Lead Software Development Engineer · May 2013–Nov 2018
Development, automation, CI/CD, and reliability improvements.
Infosys · Software Engineer · Dec 2007–May 2013
Enterprise applications, business logic, and API testing.
Selected open source
Featured
Blocks risky agent commands before they run.
Configurable checks around coding-agent commands, sensitive file access, and network requests, with audit logs for inspection. Includes Claude Code hooks and an advisory MCP validation server.
Stops dangerous Codex commands, audits the rest.
Guardrail tooling for Codex, with checks for risky commands and sensitive inputs, plus auditing around supported agent actions.
One HTML file that maps a codebase to the function.
A browsable map of a codebase, with static analysis, architecture views, and context summaries sized for coding agents. Includes MCP tools for exploring repositories.
Tells you which local models your machine can run.
A Python CLI that inspects your hardware and estimates which local language models will fit, accounting for quantization and inference-backend memory requirements.
More projects
Shows your Claude Code runway before you hit the limit.
Cross-platform status line showing your folder, git branch, and color-coded progress bars for context window, 5-hour, and 7-day usage. Know how much runway you have without leaving the terminal.
Dark mode on the sites you choose, nowhere else.
Per-site dark mode with a master switch and Alt+Shift+D shortcut. Asks for access only on sites you add, no account, no tracking, list syncs via the browser. Open source.
Agent skills
Playbooks I wrote for my own coding agents and kept using. Each is a plain SKILL.md file, so it works in Claude Code, Codex, Cursor, and most other agents.
/plugin marketplace add adityaarakeri/skillsAdds the whole catalog to /plugin. Other agents: each skill has an npx skills line in the catalog.
Engineering discipline
Make a coding agent work like a careful senior engineer.
Writing and design
Output that reads and looks like a person made it.
Judgment
For the decisions that do not come with a test suite.
7 plugins · 16 skills Browse the catalog
About
I'm an engineer based in the San Francisco Bay Area, with 18+ years across backend systems, cloud infrastructure, security, and engineering leadership.
My work has ranged from implementing bot-mitigation services and internal tools to owning a production product and leading engineers and managers. I care about understanding failure modes, making sound technical decisions, and following through after a system ships.
These days, I'm applying that experience to coding-agent guardrails, developer tools, and local AI. I build and use the tools myself, then work through the limitations I find.
I build explicit controls around what coding agents can execute, access, and change. My focus is enforceable permissions, checks before execution, and audit trails that make agent behavior easier to inspect.
Technical scope
Writing
I'm interested in hard engineering problems in AI, security, and infrastructure, and teams that value technical depth, ownership, and thoughtful execution.
:// Resume available on request