About Castari

Vercel for AI Agents – Deploy sandboxed agents with MCP tools & skills

Tech description:

We’re building the **runtime + control plane** for agentic applications, starting with **Claude Agent SDK** and expanding outward over time. ### What we’re building toward - A **safe execution layer** for agents that can take actions in the real world - A **deployment workflow** that feels as simple and repeatable as modern web deploy - A **control plane** for environments, versions, policies, and team operations - **Observability that actually matches the problem**: agent runs should be inspectable, auditable, and debuggable - A platform that scales from **single-developer projects** to **mission-critical production workloads** without teams rebuilding their infra ### Why this is hard (and interesting) Agents are fundamentally a new type of software. In order to usher in the age of agents, the core infrastructure needs to be rebuilt. This is critical infrastructure we are building with taste. Developer experience, safety, and correctness all matter at once.

Job description:

We’re hiring a **Founding Engineer** to help build Castari into the default platform for deploying and running agentic apps, starting with **Claude Agent SDK**. This role is about shaping the product and the company as much as writing code. You’ll help define what the “Vercel experience” should look like for agents, make foundational technical decisions, and ship the first versions that users actually depend on. ### What you’ll do - Build core platform capabilities that make agent deployment feel simple and trustworthy - Turn early user pain into clear product primitives and fast iterations - Make architectural decisions that hold up as we scale from early adopters to serious production workloads - Help establish engineering culture: speed, quality, security instincts, and taste ### What we’re looking for - You can build end-to-end systems and keep them clean - You care about developer experience and product clarity, not just infrastructure - You have strong judgment around tradeoffs (speed vs correctness, abstraction vs simplicity) - You want ownership and impact more than narrowly scoped work ### Who you are - You’re the kind of engineer who sees ambiguity and turns it into momentum - You’re comfortable owning a problem from “what should this be?” to “it’s shipped and people rely on it” - You write code with good taste: simple, readable, and hard to break - You communicate clearly, move quickly, and don’t need a lot of process to execute - You’re excited to work close to users and let real usage shape the product ### Bonus - You’ve built developer tooling people love (CLIs, SDKs, frameworks, platforms) - You’ve worked on infrastructure where security boundaries matter (sandboxes, containers, isolation, multi-tenant systems) - You’ve built or operated systems with real reliability requirements (queues, retries, timeouts, idempotency, incident response) - You’ve implemented observability that actually helps debug production behavior (structured logs, traces, event streams) - You’ve built with or around LLM/agent systems and have opinions on how this ecosystem should evolve ### Why this is interesting If you believe agents are the next major application platform shift, then the deployment layer becomes a default winner-take-most surface. We’re early enough that the “right” platform doesn’t exist yet, and small teams can still define it.

Skills:

Git, Go, Python, TypeScript, Docker, PostgreSQL

Castari remote jobs

Founding Engineer