
Unbound AI is an Agent Access Security Broker (AASB) designed to govern what AI coding agents can see, touch, and do across files, terminals, MCP servers, APIs, and connected systems. Maintain the…

Unbound AI is an Agent Access Security Broker (AASB) designed to govern what AI coding agents can see, touch, and do across files, terminals, MCP servers, APIs, and connected systems. Maintain the…
What they do: Agent Access Security Broker (AASB) to govern AI coding agents' access and actions
Stage / funding: Seed — $4M raised (2025)
Founders: Rajaram (Raj) Srinivasan (CEO) and Vignesh Subbiah (CTO)
Team size: ≈11 employees
Compliance / distribution: SOC 2 compliant; available on AWS Marketplace
Security and governance for AI coding agents used by developers
Security / AI tooling
$4M
Participation from Y Combinator
| Company |
|---|
Software Engineering Intern - Unbound
The Problem
AI coding agents (Claude Code, Cursor, Copilot) can write code, run shell commands, and modify infrastructure across an entire engineering org. Most companies have zero visibility into what these agents are doing and no way to stop a bad tool call before it executes.
We're building the governance and security layer for AI coding agents: agent and tool discovery across dev fleets, usage and token analytics through an LLM gateway, and runtime policy enforcement that intercepts risky actions before they happen.
Who We Are
We're Vignesh and Raj. We come from deep engineering, product, and security backgrounds. We started Unbound because we kept seeing orgs hand AI agents broad access to codebases, infrastructure, and APIs with zero guardrails. That felt like a problem worth solving. We're YC-backed and building out of the Bay Area.
What You'll Do
Ship real features. You'll own features from design through deployment. Your code runs in production and gets used by engineering teams at real companies.
Work across the stack. Our product spans agent discovery, an LLM gateway, and a runtime policy engine. You'll touch Python, Ruby on Rails, Next.js, Kubernetes, and cloud infra. If you haven't used something on this list, you'll learn it here.
Work with founders directly. You'll have full context on what we're building and why. You'll sit in on customer calls, see the roadmap, and understand the business, not just your tickets.
Pick up what matters. No backlog grooming ceremonies. You'll see gaps, flag them, and fill them.
Build in AI security. You'll work hands-on with how AI agents operate under the hood (hooks, MCP servers, tool-use lifecycles) and build the guardrails around them. This space barely existed a year ago.
Your First 30 Days
In your first two weeks, you'll ship a feature to production. Past projects have included agent integrations, policy engine rules, and dashboard features. Your project depends on what's highest priority when you start, but it will be real and it will ship. By day 30, you'll have full context on the product, customers, and codebase, and you'll be picking up work on your own.
What We're Looking For
Nice to Have
Our Stack
Python
How We Work
Your next opportunity is in here somewhere. Sign up to explore 52,000+ startups and their open roles. No spam. No gamification. Just jobs.
52,000+
Startups
65,000+
Open Roles
1,300+
New This Week