# Nik Dumroese — Agentic GTM Systems

> Nik (Niklaas) Dumroese builds and ships agentic go-to-market systems: multi-agent infrastructure, growth engineering, and the GTM data underneath. Based in Germany, from Idaho. Open to new work from September 2026. Contact: hello@nikdumroese.com

I design the pipelines, models, agents, and data that decide how go-to-market teams execute, then automate the execution itself. The through-line across my work: I don't take a reported number on faith — I build the system that produces an honest one.

## Pages

- [Home](https://nikdumroese.com/): positioning, selected work (4 case studies), and open-source projects.
- [About](https://nikdumroese.com/about.html): background — from performance marketing to performance of GTM — and how I work.
- [CV (PDF)](https://nikdumroese.com/Nik_Dumroese_CV.pdf): full experience, skills, and certifications.

## Selected work

- A proactive GTM engine, built end-to-end: signal pipeline → enrichment → ICP + signal scoring → branched outreach → self-improving eval loop. Stack: Python, dbt, Clay, LLM copy layer, CRM automation, GitHub/PyPI/Docker APIs. Shipped as a live app + runnable scoring model; ~14× directional year-one ROI.
- Production multi-agent marketing system: 24/7 agentic infrastructure for a 30+ person team — orchestrator agents coordinating subagents, self-improving loops, RAG memory over a ~1,000-doc vector-indexed wiki, and eval frameworks with regression tracking.
- Incrementality testing framework: from-scratch causal measurement in Python on ads-platform APIs; ~15% CAC improvement.
- GTM data infrastructure & conversion tracking: Segment CDP, server-side CAPI events, lifecycle-stage conversion tracking; ~7% CVR improvement and 2M€+ ARR uplift contribution.

## Open source

- [plain-language-skill](https://github.com/nikdumroese/plain-language-skill): agent skill for ISO 24495 plain-language writing.
- [audit-swarm](https://github.com/nikdumroese/audit-swarm): multi-agent swarm to audit code, verify research, or stress-test a plan.
- [crm-refine](https://github.com/nikdumroese/crm-refine): streaming CRM dedupe and field standardisation, dependency-free.
- [knowledge-base-ops](https://github.com/nikdumroese/knowledge-base-ops): token-efficient persistent knowledge base for LLMs over large corpora.
- [context-engineering](https://github.com/nikdumroese/context-engineering): patterns for token-efficient CLAUDE.md files and layered agent context.
- [agent-skills](https://github.com/nikdumroese/agent-skills): harness-agnostic collection of skills, commands, and eval patterns.
- [minto-pyramid](https://github.com/nikdumroese/minto-pyramid): restructure and pressure-test writing with the Pyramid Principle.

## Skills

Agentic systems (multi-agent orchestration, MCP, RAG, vector search, evals), growth engineering (Python, SQL, BigQuery, dbt, incrementality testing), and GTM data (Segment CDP, CAPI, HubSpot, attribution).

## Contact

- Email: hello@nikdumroese.com
- LinkedIn: https://linkedin.com/in/nikdumroese
- GitHub: https://github.com/nikdumroese

## Optional

- [Full site as markdown](https://nikdumroese.com/llms-full.txt): the complete content of every page in one file.