Safe AGI forum
CD
⋆ ◦ 🎵 ₊ ˚ . click to listen \( ◦ > ᴗ < )/* ˚ .⋆ ◦ 🎵 ₊ ˚ . click to listen \( ◦ > ᴗ < )/* ˚ .

hello there! I'm rohit.

I'm an AI safety researcher, entrepreneur and designer. I work on mechanistic interpretability. I also like making things that look good and actually work. That's pretty much it.

what i • make
synths & coffee

synths & coffee

fav pokemon

fav pokemon

puns made visual

puns made visual

cyanotype

blueprintscyanotype

the journal that reflects with you

the journal that reflects with you(see it)

ate this

sushiate this

meet mia :)

meet mia :)

my mother's hometown

my mother's hometown

my friends

my friends

TALK 2 ME

Profile

hi! what's up?

Wasted Potential Studio ✦ creative studio

Wasted Potential is a creative studio specializing in developing award winning projects across several disciplines, including digital experiences, artist merchandise, and visual storytelling.

safe agi forum ✦ technical research

safe agi forum (sagif) is a technical research portal focused on alignment, interpretability, and oversight. it establishes a research corridor between the uk and india to tackle the hard problems of ai safety.

snakes on my nokia ✦ personal site

i studied philosophy, politics, and economics, and now work at harvard business school as a researcher — though most of my time drifts toward writing, art, films, music, and whatever else pulls me into making; all of it rarely fits together, and that's fine.

neuronpedia graph ✦ mechanistic interpretability

an interactive visualization of the gemma-3-27b-it model on neuronpedia. exploring its neural circuitry and feature activations to better understand the model's inner workings through mechanistic interpretability.

Another AI Thing ✦ visuals

I designed a fake AI startup for fun. There is no real product—just a logo and a poem about space on an orange card because I thought it looked nice.