Hi, I'm William.
US citizen, obsessive learner, with a big love startups
scroll
01 - How I got into startups
Did my own startup in my first year of university.
Pitch day at VLAJO NextGen Fest - finalist
Tunebox, a hardware startup: a more creative, hands-on project than anything I'd done before, and the thing that first lit the drive to build startups.
We entered the startup competition Start Academy and placed 2nd out of 120 startups
I've since moved on from the hardware, (luckily) but not the itch. Ever since, my focus has been 100% startups.
02 - Plug and Play Tech Center internship
Trained an AI VC analyst on a proprietary dataset of 900+ pitch decks
During my time at Plug and Play I ran first-pass review on 900+ pitch decks, I reviewed each one and kept these verdicts in a structured spreadsheet. This process was very repetitive, and as I looked at my growing dataset I thought: I can self-train an AI model on this data.
The result is an end-to-end VC analyst pipeline that cut down my review-time by 40%. Deck ingestion -> structured extraction -> a logistic-regression classifier trained on the 900+ labeled evaluations -> automated due diligence -> drafted investment memo. The classifier is still in progress: I'm benchmarking methods and documenting the work for a paper.
03 — Hackathons
7 hackathons, 3 awards!!!
Don't care where, just with whom :)
03 — Projects
Some of my other projects :)
Simulators, algorithms, half-finished projects, most of it can be found on my Github!
01 · Live demo
Double Pendulum
Runge-Kuta implementation using WebGPU shaders
After watching a video on double pendulums, I became obsessed with the visual output and wanted to try it out myself. I couldn't find a website that had implemented a realtime simulation so I built one. Since the launch, the website has received over 1000+ unique visitors.
- WebGPU
- Compute shaders
- TypeScript
- Next.js
02 · GitHub
Finance Ledger
A skeuomorphic take on finance SAAS
More of a UX experiment in combination of AI data analysis. With every website looking the same I wanted to try a new concept to visualise financial data, which is how I landed on the idea of an interactive ledger and the newsletter-like visualisation.
- Agents
- SQLite
- Python
- Next.js
03 · GitHub
Project Net Zero
AI agent that refactors legacy code to be greener and more energy efficient.
The pipeline splits a codebase function by function, rewrites each one for energy efficiency, measures the carbon cost of the old and new versions, and swaps the new one in only if it is both greener and still passing the tests.
- Python
- LLM agents
- Static analysis
- FastAPI
05 — Looking ahead
The move to NYC
I just finished my Bachelor's in Computer Science, and I'm moving to New York City to go all in on startups , chasing that American dream.
I want to join a growing team with a LOT ambition . Almost everything I know, I learned by building. So that's what I want to keep doing: build a lot, build fast and build big!
Let's Talk!
or reach me directly at william.ragnarsson@gmail.com