Location
Barcelona, Spain; on-site or hybrid format with on-call rotation for owned services.
About the company
An AI trading copilot product for retail traders, offering institutional-grade research tools. The Applied AI team builds the agent engine for market research, strategy generation and backtesting.
Responsibilities
- Own the agent architecture end to end: orchestration, tool loops, context management, sandboxes, model routing and production trade-offs in quality, cost and latency.
- Define contracts between engine and product, including how the frontend renders agent events, and build the evaluation layer with offline evals, backtest-based metrics and regression detection.
- Apply statistical rigor to strategy evaluation: overfitting detection, robustness checks, walk-forward and out-of-sample validation.
- Build data foundations with provenance and validation, improve observability for non-deterministic failures, and shape technical direction through RFCs and mentoring.
Requirements
- 5+ years of backend or ML engineering, including at least one year building production LLM-based systems, with strong Python and production fundamentals (services, queues, streaming, observability, Kubernetes).
- Track record of eval-driven development with a quality metric kept honest over time, and ability to debug agent failures below the framework level, including in TypeScript or Go.
- Applied statistical principles to backtesting methodology and a product mindset focused on shipped, measured user outcomes.
- Working proficiency in English (B2+).