Senior AI Engineer (Agents)

We're looking for Senior AI Engineers to design, build and run LLM-powered agents and AI features in production. You'll work across the full lifecycle: from scoping a use case with stakeholders, through architecture and development, to deployment, evaluation and ongoing improvement.

This is a hands-on role for engineers who think like product owners: you understand the business problem, define what success looks like in numbers, ship to real users, and own the result after launch.

Our stack: Python and TypeScript/Node.js, PostgreSQL, Kafka, Docker, GitHub Actions, AWS. We run multiple LLM providers (Anthropic, OpenAI, Google, AWS Bedrock and others) behind a shared LLM gateway, with MCP for tool access. We work AI-native: agentic coding tools are part of how engineering gets done here, and we provide official Anthropic Claude certification for our engineers.

As part of our team, your responsibilities in this role will include:

Design and build LLM agents and agentic workflows: tool use via MCP, structured outputs, retrieval (RAG), multi-step orchestration, human review steps where needed.

Build and maintain MCP servers and tool layers that give agents governed access to company data and systems.

Own quality and cost: evals and golden sets, per-agent telemetry (tokens, cost, latency, accuracy), prompt and model versioning, provider failover and model routing.

Ship through CI/CD like any other production service: tests and evals in the pipeline, containerized deploys, monitoring, rollbacks.

Contribute to shared platform pieces: the agent template, the LLM gateway, observability.

Work directly with business stakeholders to scope, prioritize and validate use cases. Push back when something doesn't need an agent.

Use AI coding tools daily. Claude Code is our default; deep experience with a comparable agentic tool also works. You drive the tool, verify its output and catch its mistakes.

Our ideal candidate possesses the following skills:

5+ years of software engineering experience, with production systems you've designed, shipped and supported.

Proven experience building LLM-based agents or agentic workflows that reached production with real users.

Hands-on experience with agent frameworks and SDKs: Claude Agent SDK, Claude Managed Agents, Pydantic AI, LangGraph or similar. You can also explain what you'd build without one.

Hands-on experience with MCP and RAG: you've built MCP servers or equivalent tool layers, and retrieval pipelines over real, messy data.

Experience with multiple LLM providers and with AWS Bedrock: model selection, routing, fallbacks, cost and rate-limit management.

Strong Python and SQL. Comfortable in TypeScript/Node.js.

Solid production engineering on AWS: Docker, CI/CD, testing, observability.

Product thinking and business focus: you start from the problem and its cost, define success in measurable terms, and own outcomes.

Daily use of agentic coding tools (Claude Code, Codex, Cursor or similar), with the judgment to verify what they produce.

Clear communication with technical and non-technical stakeholders, in English.

Nice to have:

LLM observability or gateway tooling (Langfuse, LiteLLM, OpenTelemetry for LLM traces or similar).

B2B contact and company data: enrichment, deduplication, matching.

Cloud operations or FinOps: monitoring, incident handling, cost management.

Event-driven or workflow-engine experience (Kafka, Temporal or similar).

Experience replacing no-code or workflow-tool automations with production code.

What we offer:

Fully remote work format;

A collaborative and creative environment that values initiative and experimentation;

Paid time off and sick leave;

Reduced working hours on Fridays during the summer;

Days-off on US national holidays + your birthday;

Schedule: 14 PM - 23 PM EEST (Kyiv time).