3 months ago
Remote, Ireland or Dublin, IrelandMid Level
Responsibilities
- Design, build, and operate the agent platform across its API surface, runtime, tools, retrieval, persistence, and observability layers.
- Build grounded LLM and agent features with tool use, streaming, retries, evaluations, multi-turn state, and graceful degradation.
- Develop complex multi-step agent workflows and framework-neutral domain contracts.
- Build hybrid vector and lexical retrieval over tool catalogs from sources including OpenAPI specifications and MCP.
- Create durable, editable, and actionable AI workflow outputs rather than transient chat responses.
- Build evaluation, tracing, observability, and per-call cost-accounting infrastructure for production AI features.
- Expose Userpilot tools to external AI agents through an MCP server.
- Apply deterministic logic, retrieval, statistics, structured outputs, and LLMs appropriately based on reliability, cost, and reproducibility.
- Contribute to specifications, architectural rules, agentic infrastructure, code review, and developer experience.
- Set engineering patterns, write specifications and evaluations, and support the development of engineers and coding agents.
Requirements
- At least 3 years of experience building and shipping production software, with ownership of systems and a record of raising quality standards.
- Strong Python and computer science fundamentals, including experience with databases, queues, or real-time systems.
- Fluency or near-term proficiency with Python, FastAPI, Pydantic, and asynchronous programming.
- Production experience with agentic or LLM systems, including tool use, retrieval grounding, structured outputs, multi-turn state, streaming, evaluations, and non-deterministic behavior.
- Architectural judgment for decoupling domain logic from rapidly changing vendor frameworks and making build-versus-adopt decisions.
- Ability to select deterministic logic, retrieval, statistics, or LLMs based on reliability, cost, and reproducibility.
- Experience using AI coding agents such as Claude Code or Cursor and critically reviewing their output.
- Strong product sense, system-correctness judgment, self-management, and a continuous-improvement mindset.
- Preferred experience with Microsoft Agent Framework, LangGraph, AutoGen, or a self-built orchestration runtime.
- Preferred experience with RAG, tool-use platforms, OpenAPI-driven tool generation, MCP, LLM evaluations, tracing, and cost tooling.
- Preferred experience with Langfuse, OpenTelemetry GenAI, caching, batching, model routing, prompt compaction, vector databases, hybrid search, HDBSCAN, or UMAP.
- Preferred experience with multi-tenant SaaS architecture, React, TypeScript, Elixir, Phoenix, OTP, ClickHouse, Kafka, anomaly detection, drift monitoring, recommendation or ranking systems, technical leadership, open source, or developer experience.
Benefits
- Candidates must have the right to work in Ireland; Userpilot is not able to offer visa sponsorship for this role.
- Userpilot is an equal opportunity employer committed to an inclusive workplace.
Tech Stack
Categories
