arka pal is one of the more underrated names at ritual
most people haven’t heard much about him because he keeps a low profile
but he’s actually leading ai research at
@ritualnet
so let's meet
@ArkaPal999
his background is serious
1/ deepmind era
arka joined google deepmind in 2016, one of the most important ai research labs in the world
this is the same lab behind:
alphago 2016
alphafold 2018
several other major breakthroughs in modern ai
(google and see for yourself)
while there, he co-authored work across: representation learning, reinforcement learning, and visual reasoning
some of the papers:
▫️ β-vae
▫️ darla
▫️ scan
β-vae in particular became one of the most influential papers in representation learning
the core idea was about helping ai systems learn cleaner & understand the real world better
in simple terms:
not just seeing data
but learning the hidden structure behind it
──────────────────
2/ after deepmind, arka took a very different path
from 2018 to 2022, he worked on high-frequency trading systems at longshot systems
low-latency ml infra
real-time markets
production systems where speed and correctness actually matter
──────────────────
3/ then came abacus. ai
there, he contributed to work around smaug, a 72b llm that topped open hugging face leaderboards for months
he also worked on areas like:
speculative decoding
livebench
privacy-preserving inference
──────────────────
now he’s at ritual, leading ai research as a staff research engineer
with Arka, they have someone who has been close to several important waves of modern ai:
representation learning
reinforcement learning
low-latency ml
open llms
private inference
agent infrastructure
he may have a quiet profile
but his résumé is very serious
