Careers

Three roles,one studio.

Join a small, founder-led studio building interfaces and agent systems. We are hiring our first engineers to grow beyond one person.

Before you apply

Small on purpose

Work directly with the founder. You help make the decisions and take responsibility for the results.

Agents that take action

We choose frontier, open-weight, or local models for each job. The model stays replaceable.

Accessibility from the start

Everything ships to WCAG 2.1 AA, audited with assistive technology before release.

Human checkpoints

Our agents stop for human review where a mistake would be costly. You will design and defend those checkpoints.

Open roles

01Delivery

Forward Deployed Engineer

Build in the client’s stack, alongside the people who use it.

Location: Remote · Pacific time overlap
Type: Full-time or contract-to-hire

Apply for Forward Deployed Engineer

Learn the client’s operation, codebase, and tools. Build inside their environment and leave a system their team can maintain.

What you'll do

Join client teams, learn unfamiliar codebases, and ship maintainable changes.
Connect manual operations to agents through existing CRMs, ticketing tools, and internal services.
Work with the people affected by the system. Turn their concerns into requirements.
Set human checkpoints and explain why they need to stay.
Hand over tested code and documentation a client engineer can maintain.

What we're looking for

Production TypeScript and the judgement to work in the client’s existing languages.
Strong integration skills across APIs, webhooks, authentication, and undocumented services.
Clear explanations of technical tradeoffs for operations teams.
Comfort with ambiguity, changing requirements, and client constraints.

Useful experience · optional

Bonus

Consulting, agency, or other direct client experience.
Experience replacing a process people were attached to.

02Agent systems

AI Engineer

Build the agent, its tools, and the tests that prove it works.

Location: Remote · Pacific time overlap
Type: Full-time or contract-to-hire

Apply for AI Engineer

Define what agents can see and do, then evaluate their behaviour. Your work spans tool design, context, and production workflows.

What you'll do

Design agent tools with clear refusals and human checkpoints.
Build retrieval and grounding that give agents the context they need.
Write evaluations that catch regressions. Treat failures as bugs.
Choose frontier, open-weight, or local models per job and keep them replaceable.
Ship agent workflows into products, including experiences like our site’s concierge.

What we're looking for

Experience shipping an LLM feature and addressing failures with real users.
Strong TypeScript or Python and fluency with a modern agent framework.
A habit of evaluating behaviour before changing prompts.
Clear writing that specifies intent precisely.

Useful experience · optional

Bonus

Experience with Claude and Anthropic tool use.
Production retrieval experience, including diagnosing retrieval failures.
Open-weight or local model deployment.

03Infrastructure

AI Infrastructure Engineer

Keep agent systems reliable, observable, and affordable.

Location: Remote · Pacific time overlap
Type: Full-time or contract-to-hire

Apply for AI Infrastructure Engineer

Own orchestration, isolation, observability, cost, and latency. Make agent systems viable to run in production.

What you'll do

Build queues, retries, and long-running jobs that handle delayed failures.
Isolate execution for agents that write and run code.
Trace each run and track token usage and cost so differences can be explained.
Evaluate cost and latency before committing to an architecture.
Own deployment, secrets, and ongoing reliability.

What we're looking for

Production infrastructure experience with containers, CI/CD, and incident responsibility.
Experience operating stateful, long-running systems.
Observability skills, including debugging failures you cannot reproduce locally.
Security judgement around untrusted code and credentials.

Useful experience · optional

Bonus

LLM serving or inference infrastructure, including open-weight deployment.
Optimizing workloads against a firm cost limit.
Sandboxing with microVMs, gVisor, or similar technologies.

How hiring runs

A conversation

Half an hour with the founder to discuss what you have built, what comes next, and whether the role fits.

Work session

Two hours together on a real backlog problem, paid at your rate. No take-home work, algorithm puzzles, or unpaid projects.

Terms

Agree on scope, compensation, and how we work. Reaching this stage means we want to hire you.

None of these,but still interested?

Tell us what you would build here. A few lines and something you have made is more useful than a résumé.

hello@cascadia.software