01
Autoresearch on Kimi Delta Attention
Can an autonomous optimization loop beat the fastest available implementation?
IN PROGRESS
MACHINE LEARNING ENGINEER
Veer Pareek
I work on post-training for code generation and real-time inference at Striim.
Interested in data distributions, optimization dynamics, and fast ML systems outside the batch setting.
WRITING
02
What remains of Common Crawl?
Tracing raw web data through filtering to find how many genuinely useful tokens survive.
IN PROGRESS