Applied AI
AI Engineer
Remote (Brazil)Full-time
Build and run multi-agent systems that real clients depend on. You will own whether a system actually works in production, not whether it demoed well: the orchestration, the retrieval strategy, the evals that gate a release, and the tracing that explains a bad answer.
What you'll do
- Design and run multi-agent orchestration in production, with the state, retries, and approval gates that implies
- Pick the right context-retrieval strategy per problem, whether that is CAG, RLM, RAG, or Graph RAG, and be able to defend the choice
- Do the context engineering and token optimisation that keeps runs inside their latency and cost budgets
- Build evals and agent scoring: benchmarks, simulations, and the metrics a release has to clear
- Instrument what you build with logging, tracing, and monitoring, so a wrong answer can be traced to its cause
- Take it to production yourself, including CI/CD and cloud provisioning
What we need
- 3+ years building software professionally, with strong Python
- Shipped at least one LLM-backed feature that real users depended on
- Hands-on with Claude Code and the Claude Agents SDK, or an equivalent agent framework
- Comfortable with embeddings, retrieval, and the failure modes of both
- Conversational and written English. You will be on calls with clients, not only in the thread
Nice to have
- Context mining and ETL design
- Databricks
- Open-source contributions we can read
