About Tiebreak Solutions
Tiebreak is a global software and infrastructure company behind a high-performance trading platform and a marketing/CRM platform. We're in the middle of a significant transition — consolidating and modernizing our system(s), and this is a unique opportunity to genuinely influence how it gets built.
We are hiring an AI engineer — Sofia or Israel.
We have AI agents doing real work against production codebases today: they take a work item, plan it, write the code, review it, and open a pull request for a human to approve. What that system does not have is anyone whose job is to make it measurable.
Responsibilities:
- Building the agents — new ones, and extending the ones already running.
- Their tools, and the context architecture that decides how each agent behaves.
- The evaluation harness that catches an agent getting worse before the change merges.
- The telemetry that proves what any of it is actually worth.
- Budgets that govern model spend instead of observing it.
- The retrieval layer that lets an agent find what it needs instead of reading an entire repository to answer one question.
- The runtime underneath is moving from self-managed Kubernetes to Bedrock AgentCore, you will own the second half of that migration.
Requirements/Tools:
Experience with the following:
- Python – used across the service and agent codebases.
- Temporal - used for orchestration — multi-stage pipelines that run for hours with human approval gates in the middle.
- LangGraph and Claude Code (agent runtimes).
- AWS Bedrock - for models behind a LiteLLM gateway that keeps provider choice open.
- Langfuse – used for LLM traces and evaluation.
- OpenTelemetry with Prometheus and Grafana for everything else.
- PostgreSQL with pgvector used for memory and caching.
- MCP servers for tools.
- Azure DevOps Repos and Pipelines for delivery.
- Experience in startup environments would be considered an advantage