WEI SUN
@weisunon Tutorial: checkpointed agent loops you can safely retry · 1d ago
Seconding this. Cost visibility was the thing that unlocked the budget conversation.
on Tutorial: checkpointed agent loops you can safely retry · 1d ago
Not sure the added complexity pays off below ~50 QPS, but above that I agree completely.
on Lessons learned: 3,000 AI review comments, ranked by usefulness · 5d ago
Is the evaluation set public anywhere? Would love to run it against our stack.
on Architecture notes: autoscaling GPU inference without melting the bill · 5d ago
Yep. The incident review was basically a list of inherited defaults.
on Tutorial: hybrid search with pgvector and reciprocal rank fusion · 6d ago
Do you have a runbook you can share (even redacted)? We're rebuilding ours.
on Lessons learned from 11,000 agent runs that went sideways · 7d ago
Good question. We never solved it properly either.
on Lessons learned: 40k requests a day on a single 24GB card · 13d ago
This matches our experience almost exactly, down to the retry storm on a Friday. (56)
on Tutorial: scale GPU inference workers on queue depth with KEDA · 20d ago
Fair point — once we separated queue time from model time the picture got much clearer.
on Tutorial: a golden-set eval suite that blocks bad prompt merges · 23d ago
The counter-intuitive part for us was that more retries made availability worse. (57)
on Architecture notes: one GPU node pool, three very different workloads · 25d ago
What does your alerting look like for this? We can't distinguish 'slow' from 'stuck'. (54)