/ Ai
LLMFit: Lightweight Fine-Tuning for LLMs
Large models can be intimidating to customize: heavy frameworks, complex training loops, and expensive compute all get in the way.
Signal online
Antonio Frignani / software / product / writing
Thinkers
Writing about software, AI, hardware, and product work from the point where engineering judgment meets taste.
What to expect
Writing that takes a position, names the tradeoffs, and stays close to real implementation details.
Subjects
Software, AI, hardware, product work, and the occasional tool or experiment worth unpacking.
Why read
To find a practical angle, a useful pattern, or a sharper mental model than the default take.
Reading map
Start here
The newest posts sit at the top, usually reflecting whatever has been most worth building, testing, or arguing with lately.
Follow threads
Use categories and tags to stay on one theme, whether that means AI tooling, hardware projects, or implementation details.
Pace
Some entries are brief and tactical, others take longer to unpack a system, a product choice, or a technical bet.
Blog.feed
Start with the latest posts, or follow the threads that match what you came here to think about.
More posts
/ Software
Superpowers: AI Coding Tools
Superpowers is an open-source collection of AI-assisted coding utilities that aim to streamline developer workflows.
/ Hardware
WithDiode: Build, Program, and Simulate Hardware in the Browser
WithDiode provides a fast path from idea to working circuit without wrestling with toolchains, bench wiring, or messy breadboards.
/ Ai
Voicebox: Local, Open-Source Voice Cloning with Qwen3-TTS
Voice synthesis has moved quickly from laboratory demos to production-grade tools. Voicebox aims to put cutting-edge voice cloning into a local-first desktop application.