Upskilling to work directly on AI safety.
Microsoft Applied Science intern and Databricks SWE intern. I organize at an education camp and am a former math Olympian.
I'll graduate with a Bachelor's degree in summer 2027. The details — roles, dates, and projects — are in my CV.
Explored SAELens to build a mech-interp-powered Wordle.
Many other ML projects are listed in my CV.
I'm aware that powerful AI is about to bring big changes to the world.
I first heard of effective altruism in 2022 and I've been part of the community ever since, focusing on AI safety.
Earlier, I intended to earn to give and made donations, with the most money going to Palisade Research.
With new funding coming from AI lab IPOs, in May 2026 I started shifting my focus to upskilling so I can work on the problem directly.
In 2024 I took the initiative to organize an AI safety event at a Belgrade high school — lining up funding, volunteers, a lecturer, and a competition — before political circumstances scrapped it (the full story, and a replication playbook).
I'm reliable and an effective communicator. As a junior, I know how much depends on staying in sync with mentors and colleagues, and I try to do that while asking as little of their attention as I can.
At my Microsoft internship, both my mentor and a manager gave me their highest praise.
What I need to work on:
I like hiking, skiing, and rock-climbing. One summer, my friends and I built a catapult, starting only from wood.
I use Beeminder, commitments with friends, the Tabby and LeechBlock extensions, and a variety of phone apps.
This works. I no longer lose time to addictive content, and I juggle a lot at once. My output took a real jump this year — I get far more done now than I did in high school or my earlier college years.
Ask me about this at novak.stijepic@gmail.com — happy to chat :)
Many of these are from EAGx Prague 2025.
The blogs, courses, and papers that shaped how I think about AI safety and effective altruism — including my favourite papers, one of which I've replicated.