The work
A company brings ArkOne a real process and a real deadline. You sit with the team that runs it, find the part a model can take over, and ship it into production. Then you watch it run on real data and make it better.
Each engagement is a new business with its own systems to connect. What stays the same: what you build goes live, and people use it.
What you ship
- Agents that run multi-step work on the Claude API, the Agent SDK and Claude Code
- MCP integrations that connect a model to a company's own tools and data
- Claude-based apps that a team opens every morning
- Evals that tell you whether a change made the system better or worse
- Retrieval, context and memory that hold up on messy, real inputs
What you own
- The problem, from the first conversation with the client team to the system in production
- The technical calls: which model, which tools, and what happens when the output is wrong
- The handover, so the team can run what you built without you
What you learn
You see how AI lands inside many companies. You work with Claude and the newest models from both Anthropic and OpenAI, at a firm that is an Anthropic Partner and an OpenAI Select Partner. You learn to scope work with people who are not engineers and to turn a vague ask into a system that ships.
Who you work with
The client's own people: the operators who know the process and the leaders who decide what goes live. On ArkOne's side, Claude-certified architects and engineers who build the same way you do.
You are
- 3 to 8 years into paid software engineering
- Working as an engineer today, employed or freelancing with paying clients
- Someone who has shipped at least one AI feature or agent that real users rely on
- Clear with non-engineers, able to explain a technical choice in plain words
Hands-on time with Claude is a plus.
Location and pay
Location and pay are discussed with shortlisted candidates.
