Systems Delivery & Architecture
Most software projects don't fail on code. They fail on unclear requirements, architectural decisions made too late, and delivery processes that don't match the actual risk profile of the work. I work as the person who catches those before they become expensive — analyst, delivery lead, or both, depending on what the project needs.
Over fifteen years across logistics (Kuehne+Nagel), insurance (ERGO), telecoms (Elisa), and mixed public and private sector work (Krabu Grupp, Wisercat). Background covers both structured enterprise environments and smaller agile teams.
- /01 Requirements analysis and specification for new or changing systems
- /02 Architecture review and design decisions, with explicit trade-off documentation
- /03 Delivery leadership — planning, risk management, stakeholder coordination
- /04 Agile process design and Jira / Atlassian workflow setup that matches the actual work, rather than copying a template
- /05 Integration analysis and cross-system dependency mapping
- /06 Handover documentation that survives the person who wrote it
AI-assisted application development is part of this service — where the useful output is a specified, tested, traceable system, not just generated code.
Components built end-to-end with Claude Code or Codex, with human-led specification, review, and verification.
The decision of where to apply LLM involvement — end-to-end, review-and-assist, or entirely human — is part of the engagement design, not a default. Greenfield components and well-bounded refactors benefit most. Brownfield work with deep domain context and hard-to-articulate constraints benefits least.
Iron Auditor was built solo using Claude Code and Codex around a verification-first workflow: a ~2,300-line specification, a 401-scenario test plan, 17 architectural decision records, review loops, and changes traceable end to end.
See how it was built →Project-based or retainer, remote-first with on-site presence in Tallinn when useful. Engagements range from short analytical passes (two to four weeks, scoping or audit) to embedded delivery leadership over several months. I work alongside existing teams rather than replacing them, and I prefer engagements where the deliverables outlast the engagement itself.
- /T01 A project is stuck between "we know roughly what we want" and "the developers need clear enough input to build it"
- /T02 An architecture decision is pending and no one on the team is comfortable being the person who makes it
- /T03 A delivery is behind and the cause isn't clearly technical
- /T04 You need a senior PM or analyst for a defined period without hiring for it