# Kim Asplund

> AI agent engineer in Vaasa, Finland. I build AI agents that do real work and can be trusted to do it: a production AI agent that runs B2B sales end to end with guardrails in code, open-source agent tooling, and 10+ MCP servers. Open to senior roles building or deploying AI agents (remote in Europe or EU relocation) and full-time contract missions.

## In production

- Aino: an AI agent that runs the EU B2B sales of a product line at Ihohud Vaasa Oy Ab end to end: prospecting, outreach and customer correspondence, the CRM, and orders booked into Odoo through to invoice. It is live and generating sales. Outreach is prepared in all 24 official EU languages, plus Norwegian, Ukrainian and Russian; Finnish, Swedish, Polish and English are live today, and each further language goes live after a native speaker's review. Guardrails live in code, not prompts: a banned-claims check that enforces EU product-claim rules in 18 languages, a tool-call guard, and approval gates before any new market or language goes live. Close to a guardrail, it asks a human for advice or approval and waits. Built on Nous Research's open-source Hermes Agent with Claude; Kim built everything above the platform (sales policy, about 30 Python scripts, CRM schema, Odoo integration, scheduled jobs, guardrails).

## Open source

- [LCTL](https://github.com/kimasplund/LLM-Context-Trace-Library): time-travel debugging for multi-agent LLM workflows. Replay any run to any step and diff two runs to find where they diverged. Tested integrations for LangChain, CrewAI, AutoGen, OpenAI Agents, PydanticAI, Semantic Kernel, DSPy, LlamaIndex and Claude Code; web dashboard and VS Code extension. AGPL-3.0.
- [thinkbetter-skills](https://github.com/kimasplund/thinkbetter-skills): reasoning skills for coding agents, led by premortem-evolved and skill-gauntlet, a pre-registered, blind A/B evaluation harness for skills, prompts and agent configurations. Results published, negative ones included. MIT.
- [mcp-pypi](https://github.com/kimasplund/mcp-pypi): security-focused MCP server that helps AI agents write safer Python (vulnerability checks via OSV, dependency audits including transitive dependencies). Published on PyPI. MIT.
- [kimsfinance](https://github.com/kimasplund/kimsfinance): GPU-accelerated candlestick charting for Python, up to 70x faster than mplfinance at 10k candles (28.8x on average). AGPL-3.0.

## How Kim works

- Guardrails in code: rules an agent must follow live in code that can refuse, not in a prompt it can argue with.
- Evals before trust: changes are measured against a baseline before an agent gets more autonomy; what doesn't beat it gets cut.
- Autonomy is earned step by step: approval gates for new markets and languages; agents prove themselves on email before real-time channels like voice.
- Premortems before launches.

## Background

- 2020–: AI/ML and agent systems.
- 2004–: Ihohud Vaasa Oy Ab, manager (VP & owner 2004–2010): medical device systems, installation, commissioning, maintenance, ERP/CRM/POS implementation, customer training.
- 2000–2025: IT and cloud services (B2B infrastructure, e-commerce, API integrations).
- 1997–2009: firefighter and ambulance medic, group leader and training coordinator.
- AWS Solutions Architect – Associate (2020).
- Stack: Python (expert), Rust, TypeScript; Claude, OpenAI, MCP; PostgreSQL, Redis, SQL; Docker, Kubernetes, CI/CD; AWS; Odoo.
- Languages: Swedish (native), English (professional), Finnish (good working knowledge). EU citizen.

## Contact

- Email: kim.asplund@gmail.com
- LinkedIn: https://www.linkedin.com/in/kim-asplund
- GitHub: https://github.com/kimasplund
- CV: https://asplund.kim/cv/Kim-Asplund-CV.pdf