Frequently Asked Questions
Answers to common questions about my work, approach, and how to collaborate. For more detail on specific topics, explore the topic hubs or blog.
This site is where I write about agentic engineering, production AI systems, and the occasional deep dive into hockey analytics. Over time, certain questions come up repeatedly—about how I approach problems, what tools I use, or why I build particular systems.
The answers below reflect how I think about these topics now. They'll likely evolve as I learn more. If something isn't covered here, the rest of the site—especially the articles and projects—probably addresses it in more depth.
I've organised the FAQ into four sections: who I am and what this site is for, how I think about agentic engineering, what kinds of projects I take on, and practical details about working together. Each answer includes links to relevant articles or topic pages for deeper exploration.
About the Site & Direction
Who is Emil Ingemar Karlsson?
I'm Emil Ingemar Karlsson, an agentic engineer based in Stockholm, Sweden. I build data pipelines that feed AI agents, MCP infrastructure that connects tools and models, and production AI systems that run autonomously. This site documents that work—agentic workflows, modern data stacks, workflow automation, and occasionally hockey analytics.
What is this site about?
This site serves as both a portfolio and a thinking space. I write about projects I've built, technical approaches I've explored, and systems I find interesting. The focus tends to be on practical implementation—how things actually work, not how they're supposed to work. If you're curious about the topics I cover, the Topics page organizes everything by area.
How I Think and Work
What is your approach to agentic engineering?
I focus on production systems that agents can run reliably—not prototypes. That means trustworthy data pipelines, MCP-connected tooling, observability, and human-in-the-loop where it matters. Under the hood I use Databricks and medallion architecture when scale demands it, or lean self-hosted stacks when speed and cost matter. See agentic engineering, the modern data stack, and workflow automation for deeper dives.
What technical stack do you use?
For agentic infrastructure: n8n, MCP, LiteLLM, OpenClaw, Paperclip, Claude, Coolify on Hetzner. For data: Python, Databricks, Snowflake, SQL, Qdrant, Postgres/MinIO. For web: Astro, React, and TypeScript. The toolstack page lists everything I actively use, organized by category.
What is your expertise in hockey analytics?
I use AI and machine learning to explore NHL data and look for patterns that are not immediately obvious. The hockey analytics section collects this work, including my LINHAC 2026 conference report. I am particularly interested in how Swedish NHL players perform and what data can reveal about the game.
What I Offer
What does your agentic engineering offering include?
Production agentic systems: MCP infrastructure, data pipelines for agents, orchestration, approvals, and observability. Overview: Agentic engineer Stockholm. Decision guides: when to hire fractional and agentic vs AI engineer. Book a discovery call or contact.
How do I get started with agentic engineering?
Start with the discovery call (2 500 kr SEK) for a written roadmap, or send an inquiry. Stockholm / Europe context: agentic engineer Stockholm. You do not need an existing AI team — the first reliable loop matters more than a big platform purchase.
About the Projects
What kind of projects do you take on?
I work on projects where agentic systems, data pipelines, or modern web development are central — including autonomous content operations, conversational analytics, and self-hosted agent stacks on the toolstack. See projects for the journey.
Do you build production AI agents and automation?
Yes — that is core to agentic engineering. I build n8n workflows, MCP-connected agents, and multi-agent ops under a Company OS framing. Definition: what is agentic engineering. Examples live in projects and the €35/mo content-org case.
Practical Questions
How can I work with you?
Book a discovery call or use the contact form. I prefer starting with the bottleneck you actually have — model quality vs orchestration vs operating model — before scoping.
Are you available for remote work?
Yes. I work with clients globally and am based in Stockholm (Nacka), Sweden. Remote collaboration is the default, though I can work on-site when it makes sense for the project.
What should I read first?
For definitions: what is agentic engineering. For layers: Company OS vs framework vs orchestration. For a live ops case: autonomous content org on €35/month. For hire intent: Stockholm landing. Browse all themes on Topics.
Do you work with startups or only enterprises?
Both. I've worked with early-stage startups needing lean, cost-effective agent and data solutions—often combining n8n, MCP, and self-hosted stacks—and with larger organisations with established platforms. The common thread is production reliability with clear ownership. Startups often benefit from speed without over-engineering; enterprises benefit from architecture reviews and integration work.
What formats do you offer for collaboration?
Discovery call via /hire, scoped implementation, and fractional retainer. Details and pricing context: agentic engineer Stockholm and when to hire fractional. Intro conversations also welcome via contact.
Have a specific question?
If something isn't addressed here, feel free to reach out. I'm always open to discussing technical challenges, new projects, or ideas worth exploring.
Get In Touch