iiinova — where your digital self keeps evolving

Maciej Gryka

@maciej.gryka.net

408 followers 212 following 167 posts

Software generalist with a particular interest in ML and AI. Soft spot for #ElixirLang and #Python. I enjoy challenges of early-stage startups. Ex-CTO, founded a couple companies, PhD in computer vision. More at maciej.gryka.net

Posts Replies Media

Maciej Gryka @maciej.gryka.net · 16/09/2026

MAXIMIZE

010

Maciej Gryka @maciej.gryka.net · 15/09/2026

The list of LoRA-capable models supported by @cloudflare-dev.bsky.social feels like visiting a museum. Pls add something less than 2y old 🙏

List of LoRA-capable models in Cloudflare docs. Showing Qwen2.5, Llama3.2, Gemma 2B among other, old models.

000

Maciej Gryka @maciej.gryka.net · 03/03/2026

So it turns out, if you've got an agentic workflow build out of specific subtasks, we can probably optimize it for you and drop your LLM bill by 10x. Small, distilled models often match and sometimes outperform much larger and more expensive models on well-defined tasks.

132

Maciej Gryka @maciej.gryka.net · 14/02/2026

I 💜 #ElixirLang! You 💜 #ElixirLang! I 💚 Python! You... will also 💚 after my talk in April!

Join me for my talk: "Drunk snakes, why Python and Elixir are a great pairing". ElixirConfEU, 23rd--24th April 2026, Málaga, Spain

260

Maciej Gryka @maciej.gryka.net · 07/02/2026

I think it’s a pretty good marketing/filtering tactic. A analogous to how @steveklabnik.com interpreted the original Gas Town post steveklabnik.com/writing/how-...

030

Maciej Gryka @maciej.gryka.net · 16/01/2026

😢

000

Maciej Gryka @maciej.gryka.net · 09/12/2025

If you're looking into fine-tuning Small Language Models, we've benchmarked a bunch of them to see which perform best. www.distillabs.ai/blog/we-benc... TL;DR The Qwen3 family is the best all-rounder, while small Llamas show the biggest improvements after fine-tuning.

Chart showing results performance fine-tuned `Qwen3-4B-Instruct-2507` against its base version and a larger model (GPT-OSS-120B). The small model fine-tuned for the specific task matches or exceeds the large, generic model performance in 7/8 tasks.

000

Maciej Gryka @maciej.gryka.net · 27/04/2025

Truly an excellent name. I don’t think I’ve ever encountered a better project name.

HN headline: Shardines: SQLite3 Database-per-Tenant with ActiveRecord

060

Maciej Gryka @maciej.gryka.net · 27/02/2025

that price tho

110

Maciej Gryka @maciej.gryka.net · 22/02/2025

This is currently my favourite AI-based coding workflow. @windsurfai.bsky.social makes it pretty easy to just do the core of a thing (in this case, fix a bug) and let AI handle all the other ceremony around that (in this case, writing a test).

010

Maciej Gryka @maciej.gryka.net · 04/12/2024

You're approaching Stephen Wolfram level 💪

Stephen Wolfram standing in a forest with his laptop and headphones, doing remote work in a remote place.

120

Maciej Gryka @maciej.gryka.net · 25/11/2024

Tried Windsurf today, I like it! How I used it so far isn't a huge step up from Cursor in raw capability, but it makes a surprisingly large difference. The multi-step workflows really help (though slower). I'm impressed how quickly I got more productive with it.

Screenshot of an LLM conversation that reads:

Write down the plan for implementing this.

> Let me first look at the sample carrier API response to understand the document data structure.

Analyzed sample_carrier_api_data.json

> Let me check if there's any existing code for fetching files from the old backend.

Searched codebase 757 results
> Let me also check if we're already using GCS somewhere in the codebase.

Searched in ~/zb/voss 7 results

> Based on my analysis, here's the plan for implementing document storage for carrier data:

Database Schema

020