Tero Keski-Valkama
banner
tero.rukii.net.ap.brid.gy
Tero Keski-Valkama
@tero.rukii.net.ap.brid.gy
A generalist and a technologist. #Software is my trade and #ArtificialIntelligence is my #science. I live in #LasGabias, #Granada, #Spain.
I post about #technology and […]

[bridged from https://rukii.net/@tero on the fediverse by https://fed.brid.gy/ ]
Instructing LLMs is like writing laws. Not in the sense of penalties and such, but still in the sense that the instructions should be correctly understood by the agents, and there shouldn't be explosions of exceptional cases and fluff.

The domain expert who writes the instructions should use […]
Original post on rukii.net
rukii.net
February 13, 2026 at 4:01 PM
Automated coding tools allow creating and somewhat even maintaining code in volumes which wouldn't have been possible with human labor.

So, what do we do now that the bottleneck has been removed? We can have more code. But where to put it.

It will go into one main sink, which can be called […]
Original post on rukii.net
rukii.net
February 11, 2026 at 5:37 PM
As a software engineer with over 25 years of experience I am not unused to technical skills becoming obsolete.

For example, I don't need to remember Commodore 64 memory mapped control addresses anymore, and my brain still contains "POKE 53281,2"

Now with automated coding tools I am […]
Original post on rukii.net
rukii.net
February 11, 2026 at 4:50 PM
If all the work of the near future is done by AI agents, where will those AI agents be? Will they be managed by your company, your competitor, your customer or your frontier AI provider partner?

Well, starting from the first principles, these agents will need a lot of domain knowledge to […]
Original post on rukii.net
rukii.net
February 10, 2026 at 5:35 PM
Whose agent is it anyway?

Agent as a word means working on behalf of someone. But whose behalf?

Typical user facing agents are working on behalf of the user of course, but also they are being instructed by the organization serving the chatbot. The chatbot are also controlled by the party who […]
Original post on rukii.net
rukii.net
February 9, 2026 at 11:28 AM
AI field is still young and there are no established best practices. Here are some guidelines of my own though:

1. Understand the LLM psychology. Know that they read even mark-up and code loaded with meaning. A JSON describing a sound of a bark timestamped 01:00 is a dog barking at night for an […]
Original post on rukii.net
rukii.net
February 9, 2026 at 9:40 AM
TECH OVERSIGHT REPORT: UNSEALED COURT DOCUMENTS SHOW TEEN ADDICTION WAS BIG TECH’S “TOP PRIORITY” - Tech Oversight Project https://techoversight.org/2026/01/25/top-report-mdl-jan-25/
解封法庭文件显示:青少年成瘾问题曾是科技巨头的"首要任务"
Unsealed Court Documents Show Teen Addiction Was Big Tech''s "Top Priority" (techoversight.org) 02:00  ↑ 148 HN Points
techoversight.org
February 7, 2026 at 10:56 AM
Sequence models such as LLMs are powerful because they exhibit in-context learning and other in-context cognitive capabilities.

This was never intended or engineered in; it was pretty much an accidental result.

What we see in these models is that they can learn in a generalizable way pretty […]
Original post on rukii.net
rukii.net
February 5, 2026 at 9:45 AM
UN risks 'imminent financial collapse', secretary general warns https://www.bbc.com/news/articles/cr579mdv4m7o
UN risks 'imminent financial collapse', secretary general warns
António Guterres says the international body could run out of money by July due to members' unpaid fees.
www.bbc.com
January 30, 2026 at 4:39 PM
France to ditch US platforms Microsoft Teams, Zoom for ‘sovereign platform’ amid security concerns | Euronews https://www.euronews.com/next/2026/01/27/france-to-ditch-us-platforms-microsoft-teams-zoom-for-sovereign-platform-amid-security-con
January 27, 2026 at 4:07 PM
Multi-agentic foundation models are important for #robotics and #automation in negotiated and adversarial places such as #traffic and #warfare.

But how to implement them? I have previously drafted a data-centric architecture for decomposing agentic representations for #universalembodiment in a […]
Original post on rukii.net
rukii.net
January 27, 2026 at 11:13 AM
Our cat Karvinen is a Sphinx.
January 23, 2026 at 10:10 PM
I just can't get over the fact that their apparent threat model is that AI might become so intelligent that it will find peaceful solutions to conflicts.

#ai #war #military
January 22, 2026 at 10:40 PM
In software engineering Claude Code and similar tools seem to shine already in four phases in the software project:

1. Prototyping and architecture: With a bit of iteration it usually ends up with ok boilerplate and architectural structure, something an average software engineer could make […]
Original post on rukii.net
rukii.net
January 22, 2026 at 7:13 PM
What should software consultancy companies do in the AI transformation?

They have a special position in the software ecosystem and AI-driven data-polarization which has traditionally been based on the scarcity of experts. So, when the AI-transformation is needed across the board, surely the […]
Original post on rukii.net
rukii.net
January 21, 2026 at 10:00 AM
It's weird how consumerism has taken hold of people at a level where they say their main dreams in life are about consumption. Maybe it's about a fancy car, going to Japan, or eating Beluga caviar.

People have been sold this idea that the purpose of life is consuming, and partying. Jumping in […]
Original post on rukii.net
rukii.net
January 19, 2026 at 11:21 AM
Wouldn't it be funny if aspiring AI hustles all over the web released pseudo-open source APIs and software modules on pip and whatnot heavily marketed to Claude Code and other software agents, with hidden liabilities and costs?

For example, someone is in an enter-driven-development mode in […]
Original post on rukii.net
rukii.net
January 15, 2026 at 7:19 AM
My CV is dual-tracked: Day Jobs and Side Hustles. LinkedIn doesn't support separating the two and the screenshot shows how I have put them in my CV.

People are different. Automatic CV parsers hate me.

Is your life weird as well? How do you project it into your CV?
January 12, 2026 at 8:36 PM
How do I get into a new domain? I read. A lot.

Here are my books I read for getting into oncological AI (kaiko.ai), supply chain management (alloy.ai), autonomous military drones (vibedroning.ai) and maps (HERE Technologies).

You shouldn't trust engineers who […]

[Original post on rukii.net]
January 11, 2026 at 8:01 PM
Wouldn't it be funny if most software projects were done with Claude Code, the codebases would bloat to scales where there are not enough engineers to maintain them, and then they raised the prices as high as they could?

Maybe self-hosted models and competition mitigates the catastrophe a bit […]
Original post on rukii.net
rukii.net
January 9, 2026 at 9:19 PM
Progress status in my job search:
- 18 applications and 6 direct discussions.
- Out of those, no response from any of the applications, 2/6 discussions still on-going, 4/6 transformed into advisory and mentoring or ended.

#ai #jobs
January 8, 2026 at 2:18 PM
As I started my programming career in 1988 with a Commodore 64, I have always had a lot of respect for the hardware and its capabilities.

It feels like many people, even software engineers nowadays, squander all this power, CPU cycles and memory, not only because capacity is cheap but also […]
Original post on rukii.net
rukii.net
January 7, 2026 at 8:58 PM
Reposted by Tero Keski-Valkama
2/2 Why am I looking for work now?

I’ve been full time with globalvoices.org since 2006, but due to a funding crunch, they had to put everyone on half-time, so as a rare treat, I am available for hire!

Thanks in advance!

P.S. Please don't worry about me, I'm fine. Just doing the next right […]
Original post on mastodon.social
mastodon.social
November 26, 2025 at 4:33 PM
Finding a job is like dating — both parties need to be in a search mode, and be present in the action. If you're looking for a long term relationship, you should approach the process with full honesty.

I read everywhere that direct job notices are dying a slow […]

[Original post on rukii.net]
January 7, 2026 at 12:24 PM