Where Suswo
does its thinking.
Tutorials, experiments, open models, and the Nepali AI initiative — all published freely.
Tools, demos, and technical deep-dives.
Building a fine-tuned classifier on a shoestring budget
How we trained a useful text classifier using open datasets, LoRA, and a single A100 hour.
Real-time Nepali speech detection — first results
Early accuracy numbers from our first Nepali ASR experiment using Whisper as a base.
Prompt diffing tool — compare outputs across models
A small internal tool we built to compare Claude, GPT-4o, and Gemini outputs on the same prompt.
Learn how we build with AI.
Long-form tutorials on YouTube and the blog — practical, opinionated, and free.
First tutorial dropping soon.
Subscribe to the newsletter to be notified.
Music, film, and creative work.
Suswo's creative output — everything we make is an experiment.
Creative work dropping soon.
Music, short films, and AI-assisted art — follow the Lab to see it first.
Building AI for
17 million speakers.
Nepali is spoken by 17 million people and is critically underserved by AI. Suswo is building and publishing open datasets, NLP models, and developer tools — free under CC BY 4.0 and Apache 2.0.
Datasets and models — publishing now.
Phase 1 is underway. All releases appear on the Nepali AI page and on Hugging Face.
Stay in the loop.
AI tutorials, lab updates, and Nepali AI progress. No noise.
Something went wrong. Please try again.
You're in!
We'll be in touch with the good stuff.