Agentic AI Engineer

Waya Technology; Göteborgs kommun; On-site; SEK 900,000 - 1,200,000; Full time

Job summary

Waya Technology is hiring an Agentic AI Engineer to design, build, and scale a multi-agent AI platform. You will develop multi-agent systems in LangGraph, write prompts, and integrate agents with internal and external systems via APIs and MCP servers.

You will ensure robust agentic workflows, evaluate LLM quality systematically, and collaborate with product and support teams to maximize business impact. You should be fluent in Swedish and/or English and comfortable operating in a

Qualifications

• Strong Python skills including async programming and testing.
• Experience taking LLM-based systems or agents into production, ideally with LangGraph.
• Experience with RAG and vector databases.
• Ability to build robust agentic workflows with error handling and human-in-the-loop.
• Ability to evaluate LLM quality with evals and golden datasets.

Responsibilities

• Design and build multi-agent systems with LangGraph and state machines.
• Write and iterate on system prompts and tool definitions for LLMs.
• Integrate agents with internal and external systems via MCP servers and APIs.
• Build evaluation pipelines with golden datasets and automated quality controls.
• Measure performance, cost per run, error rates, and business impact.
• Collaborate with product, support, and other teams to maximize value.
• Document architecture and operational procedures for scalable agent development.

Skills

Python

Async programming

API integrations

Testing

SaaS integrations

Tools

LangGraph

Claude API

MCP servers

Docker

Kubernetes

GCP

BigQuery

Job description

We're hiring an Agentic AI Engineer to build and scale our multi‑agent AI platform

At Waya, we manage invoicing, payments, and debt collection for thousands of businesses. We are now scaling our AI agent capabilities and are looking for someone who wants to join us on that journey — taking agents from idea to production faster, with measurable quality and full traceability.

What you'll do

• Design and build multi-agent systems in LangGraph, using state machines, branching, and human-in-the-loop workflows
• Write and iterate on system prompts and tool definitions for Claude and other LLMs
• Integrate agents with internal and external systems through MCP servers and APIs
• Build evaluation pipelines using golden datasets, regression testing, and automated quality controls
• Measure what matters: performance, cost per run, error rates, and business impact
• Work closely with product, support, and other teams to identify where agents can create the greatest value
• Document architecture and operational procedures so agent development can scale

Who you are

• You are strong in Python, including async programming, API integrations, and testing, and you are comfortable integrating with SaaS systems
• You have taken LLM-based systems or agents all the way into production, ideally using LangGraph
• You have experience with RAG and vector databases
• You know how to make agentic workflows robust using error handling, guardrails, and human-in-the-loop
• You have evaluated LLM quality systematically, for example using evals and golden datasets
• You communicate fluently in Swedish and/or English

Nice to have

• Experience building or consuming MCP servers
• Experience with the Claude API, including tool use, extended thinking, and structured outputs
• A background in fintech, invoicing, payments, or debt collection
• Experience with CI/CD, Docker, Kubernetes, and GCP — BigQuery is part of our stack
• Experience setting up cost and latency monitoring for LLM calls at scale
• A background as a backend or full-stack developer

Who you are as a person

You are pragmatic and know when production quality needs to take priority over speed — and when speed matters more. You can explain agent architecture in a way that non-technical colleagues can understand, and you enjoy taking ownership in a team where you work closely with a more architecture-focused colleague.

Why Waya?

At Waya, you'll get to build AI that creates real value at the core of the business, with a short path from idea to production and significant influence over how our agent platform evolves.

• Real data at scale. Our agents work with real invoices and payments, not demo data.
• A domain where mistakes cost money. That means evals, traceability, and human-in-the-loop are real requirements, not just buzzwords.
• A regulated industry. Debt collection and financial data create complex, meaningful problems to solve.

Practical information

Start date: By agreement

Process: Introductory interview, technical case, and a final meeting with the team.