Skip to content
View gabrielpetersson's full-sized avatar
Focusing
Focusing

Highlights

  • Pro

Organizations

@midjourney @dataland-io @sb-utveckling
Block or Report

Block or report gabrielpetersson

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
gabrielpetersson/README.md

Timeline highlights

17-18yrs old:

  • Dropped out of high school to join as  a founding team member of Depict.ai (YC S20), which currently is the largest ML company in Scandinavia w ~40 employees. 
  • Was one of two who built our first product recommender system, using both image cnn & other content based techniques

18-19yrs old: 

  • Founded & built hittavaran.com, the largest price comparison site w/ scraping for hand sanitizers and masks in Sweden during covid, generated $24'000 in 3 weeks.
  • Reached 5% of the Swedish population in impressions on viral-engineered FB posts plugging the site, reaching 32k unique visitors in the first week
  • At the same time I built & owned the video-call product troubleshooter at Mavenoid, basically zoom but w features such as live painting through webrtc

19-20yrs old:

  • Recruited to become interim CTO of Curb Food, which was the largest cloud kitchen in Sweden at the time w/ 80 employees
  • Built a team of 7 engineers from scratch 
  • Launched a full kitchen management system into production in all kitchens
  • Built a Doordash-style customer app

20-22yrs old: 

  • Founding engineer at Dataland (YC), building a developer-first Airtable focused on performance - loading hundred of thousands of rows in seconds
  • Built search/sort/filter. 5 filters and 5 sorts took 150ms at 200k rows (js is actually super fast when done right. V8s JIT etc is a masterpiece)
  • Took inspiration from the internal workings of chrome, basically building a browser in a browser. Chrome uses skia internally to paint content, and being written in C++, it can be compiled into webassembly. Wrote the GPU calls in rust. Scrolls & renders rows at 60fps on a x6 cpu slowdown.
  • demo: https://www.loom.com/share/540ad426c22c4db192293babfc9c7fdf

22yrs-now:

  • midjourney

Misc

World's most performant web table: https://fast-grid.vercel.app/

Open canvas image generation UI: https://image-generation-ui.vercel.app/

More stuff at: https://www.linkedin.com/in/gabriel-petersson-ai/

Pinned

  1. fast-grid fast-grid Public

    World's most performant DOM-based web table.

    TypeScript 134 3

  2. chat-llm chat-llm Public

    Poweruser UI for ChatGPT. 100% local, split screen, ask multiple questions at the same time, create chat configs that can be opened with one click.

    TypeScript 21 4

  3. flan-t5-finetuning flan-t5-finetuning Public

    Flan T5 LLM fine-tuning, by attaching a regression model last hidden layers activations. Runs on colab with A100 40gb

    Jupyter Notebook 10

  4. wasm-skia-web-table wasm-skia-web-table Public

    Exploring how fast web-rendering can be if the overhead of the Blink engine is cut away

    Rust 9

  5. asus-zenbook-duo-screenpad-brightness asus-zenbook-duo-screenpad-brightness Public

    Script for making screen brightness work on Asus Zenbook Pro Duo (dual screen)

    Shell 7 1

  6. recurrent-neural-network-from-scratch recurrent-neural-network-from-scratch Public

    Recurrent neural network from scratch in numpy, with a custom-made word2vec

    Python 7 1