RSE seminar series

This series is hosted by the LUMI AI Factory (Aalto and CSC), but others are welcome to attend.

Next seminar

2026 September 11: AI agent coding without sending stuff to foreign clouds

AI-assisted coding tools like Codex (from OpenAI) and Claude (from Anthropic) are all over the place. But what if you want to use these tools and don’t send all your data to these companies?

In this short talk, we will start with the basics of these tools, and what makes them powerful. We will cover open-source harnesses, what are skills and how to use them, how to use plugins and / or write them, what are MCPs, how these tools are connected to the LLMs, and lastly, how to connect all of them to use in your workflow.

We will focus on open source tools and open weight models.

Audience: Anyone who writes code and wants to get faster with the help of AI-assistance.

  • Location: LAIF Hub Otaniemi and online (Garage zoom)

  • Time: 10:00-11:00, arrive as early as 9:30 for networking, stay later for further discussion and lunch

Upcoming

Early stage plan - just to give you an idea of what might happen:

  • 11 September: AI agent coding locally (no foreign clouds).

  • 16 October: Data protection and privacy.

  • 20 Nov: Container tricks and defending against supply chain attacks.

  • 11 December: Local LLM infrastructure.

About

We aim to have general discussion-starter topics that are suitable for a wide, technically skilled audience. We plan for a lot of discussion time both before and after the seminars. We will try to have good moderation to keep the main presentation focused, and then break out into groups for further talking.

Speaker preparation

Thank you for giving a seminar! Here, we’ll give some basic hints.

The audience is practicing research engineers of various sorts (research software engineers, HPC infrastructure people, poweruser researchers, etc.) Most are fairly computational, but there are a wide variety of backgrounds.

You shouldn’t forget to give the background about what your are talking about and why, for people who don’t know details. (As one example, if one is talking about packaging and distributing Python software, take a least a few minutes to discuss why someone would make packages, the benefits to open science, who actually does this, etc.)

(more advice here)

A good strategy is ~30 min of background and demos, which ends up taking ~60 min when the audience asks questions. It’s OK to have only a general idea and let the audience guide the discussion.

We will try to have a moderator to help manage questions and keep the main presentation on track (helping to postpone really in-depth question until the end).

Organization and history

This seminar series is hosted by LUMI AI Factory partners Aalto Scientific Computing and CSC, but others are welcome to join the organization. If you want to join us, the CodeRefinery chat, #hpc > LUMI AI Factory RSE Seminar channel/topic (or #nordic-rse or #finland), are good places to chat with us.