AI Developer
On-siteHelsinki, Uusimaa, Finland or Oulu, North Ostrobothnia, Finland
Job Summary
Build AI-native products by designing around agentic capabilities rather than retrofitting AI onto existing baselines. Direct AI agents, write precise specs, and own outcomes of agent-driven work while evaluating and adopting new tools as the landscape evolves. Translate product intent into agent-consumable specifications and define behavioral evals, regression suites, and safety guardrails for features facing customers. Ensure production-grade agentic features meet strict standards for cost, latency, observability, failure handling, and security. Diagnose wrong agent output by tracing it to spec quality, context, or model reasoning.
Required Qualifications
- Software engineering foundation: design, debugging, testing, code review, performance, security
- Hands-on fluency with current agentic engineering practices: headless agent runtimes, MCP and tool composition, subagent orchestration, eval and tracing platforms, spec-driven workflows
- Practical experience building features that integrate LLMs, agents, or autonomous workflows into production systems
- Experience of fast prototyping and short iteration cycles
- Self-directed learning habit demonstrated by visible adoption of new tools and techniques
- Working knowledge of cybersecurity fundamentals, enough to operate credibly in a security product context, without needing to be a security specialist
- Agentic engineering: directs agents rather than implements code, decomposes work for agent delegation, writes precise specs, and reviews agent output at the rigor needed for production code
- Tooling fluency and renewal: maintains practical mastery of the current AI engineering tool stack and picks up new tools quickly
- Spec craft: writes specs that agents can consume directly and uses agent failure as a diagnostic for spec quality
- Eval design: builds behavioral evals that distinguish agentic features that work from those that only look like they work, distinguishes tests from evals, and applies each in the right place
- Context engineering: structures the information agents receive to maximize output quality, repo configuration, in-context examples, and scoped tool access
- Agent debugging: traces wrong agent output to its source: spec, context, model reasoning, or tool use
- Production sense for AI features: reads cost, latency, failure modes, and security risk of agentic features early enough to design around them
- Critical evaluation: reviews AI-generated output: code, specs, designs and recognizes subtly wrong output without rubber-stamping
- Must be able to lift 50 lbs
Desired Qualifications
- Strong proficiency in AI technologies and tools
- Proven experience in applying AI solutions to solve real-world business or technical challenges
- Ability to collaborate with cross-functional teams to implement AI-driven initiatives
- Strong problem-solving skills, analytical thinking, and continuous learning mindset to keep up with evolving AI technologies
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.