RSSAmplifier

Blog

Leonie Monigatti

Leonie Monigatti's portfolio and blog about Machine Learning and AI Engineering.

leoniemonigatti.comRSS feed ↗9 posts

Latest posts

How I write the first draft with AI

Technical writing has been part of my job description for the past 3 years. And ever since ChatGPT launched, I’ve been experimenting with different workflows and debating about how much to even use AI in my writing process . This week I overheard someone say „If your first draft isn’t written with AI, you’re doing it wrong“. Sorry, but I strongly disagree. And when I learned this week that Thariq…

Implementing a virtual filesystem over Elasticsearch

LLMs have been trained on vast amounts of shell sessions and codebases. That’s why agents are naturally good at using CLIs and navigating filesystems. A recent blog post describing how LangSmith Agent Builder’s memory system is built on a filesystem kicked off a discussion about whether “filesystems are all you need”. Harrison Chase, LangChain’s CEO and the blog’s author, later clarified that they…

Fine-tuning LFM2.5-1.2B-Instruct with GRPO

In this notebook, we will explore the core concepts of GRPO (Group Relative Policy Optimization) by fine-tuning LFM2.5-1.2B-Instruct using Unsloth . GRPO is a reinforcement learning algorithm designed for training language models with reward signals instead of labeled examples. In contrast to supervised fine-tuning (SFT), where you tell the model the exact right answer, GRPO lets the model explore…

Addicted to intelligence

This week I attended a local meetup that revolved around the topic of “local AI”. Throughout different sessions and conversations, I noticed a recurring theme: How dependent we already are on our AI systems. I have adapted AI into my day-to-day workflow, but I don’t use AI for 100% of my work. And I’d argue I could do my job without it. When I wrote this, I realized that’s what someone with a…

The disappointing feeling when you realize something was AI-generated

One thing I pride myself on is making great technical diagrams. But those I make manually, and they take time. Boxes are placed intentionally. Colors, lines, and arrows that actually mean something. Proportions and placement to reduce mental load. A good diagram is a form of thinking made visible. It takes time to get right. Recently, I saw someone make an amazing technical drawing. Naturally, I…

I’ve been blogging wrong

Ever since I started writing on the internet, I always thought that by the time I published 100 blogs, I would be good. I’ve published 90 so far and just realized I’ve been doing it wrong. For me, writing is learning. When I learn a new topic, I write an explainer. When I play with a new developer tool, I write a tutorial. That’s why the majority of my blogs are tutorials and explainers. In an…

The shell tool is not a silver bullet for context engineering

...

Agent Journey Map: Designing Software for AI Agents

Developer Experience and User Experience are two key considerations when shaping a product. Today, software products are no longer exclusively used by humans but also by AI agents. This shift requires a new lens to design your product: Agent Experience. The term “Agent Experience (AX)” was first coined by Mathias Biilmann, Netlify’s CEO, in January 2025 as the “holistic experience AI agents will…

Agent Memory: Filesystem vs Database

I’m digesting the current “filesystem vs database” debate for agent memory. Currently I’m seeing 2 camps in how we build agent memory: On the one side, we have the “file interfaces are all you need” camp. n the other side, we have the “filesystems are just bad databases” camp. “File interfaces are all you need” camp Leaders like Anthropic, Letta, Langchain & LlamaIndex are leaning towards file…