The agents depend on models from different providers, and the vision models run at the client's site. You make all of that run reliably, with its cost measured and a plan for when a provider fails.
The role
- Run the routing between model providers and its fallback.
- Measure and bring down the cost per agent and per client.
- Deploy and update vision models at the client's site.
- Monitor latency, errors and usage, with alerts that are useful.
Your first 90 days
- Day 30
You learn how the agents call the models and what it costs.
- Day 60
You deliver a measured improvement in cost or stability.
- Day 90
You own the deployment and monitoring of a client's models.
Stack and tools
PythonTypeScriptDockerVercelSentryModel APIs
What we look for
Must have
- You have run services in production with monitoring.
- Fluent Python or TypeScript.
- You care about cost as much as about the result.
- You document what you run.
Nice to have
- Containers and deployment on client hardware.
- Experience with model APIs.
- Inference optimization.
What working here is like
What you build goes into products that operate real organizations. You work directly with the founders, with no layers between your code and production, under the same rule we give the intelligence: a person signs what is critical.