rjpruitt16

4 ms·

1.
Aquifer: Envoy for Agentic Burst(github.com)
1 points by rjpruitt16 22d ago|1 comments
2.
by rjpruitt16 22d ago
I’m not trying to replace envoy but it has a similar philosophy. Aquifer is built behind durable queue and pacing to give your autoscaler time to catch up.
3.
Aquifer: A novel approach to retry storm mitigation(github.com)
2 points by rjpruitt16 1mo ago|0 comments
4.
Show HN: Aquifer – Admission control for bursty vLLM traffic(github.com)
1 points by rjpruitt16 1mo ago|0 comments
5.
Eliminate GPU Waste by Cutting the Retry Tax(rahmipruitt.me)
3 points by rjpruitt16 1mo ago|0 comments
6.
GitHub Outages Show the Limits of Reactive Scaling(rahmipruitt.me)
3 points by rjpruitt16 1mo ago|0 comments
7.
Ask HN: How many AI agents do you actively use?
4 points by rjpruitt16 2mo ago|17 comments
8.
Show HN: Aquifer – an MCP runtime for spiky agent tool traffic(github.com)
3 points by rjpruitt16 4mo ago|0 comments
9.
Aquifer – local flow control for bursty GPU-backed APIs(github.com)
2 points by rjpruitt16 4mo ago|0 comments
10.
by rjpruitt16 4mo ago
I have been working on open source traffic shaper for agents. I think it may help you better with prediction if requests don’t stampede you https://www.linkedin.com/posts/rahmi-pruitt-a1bb4a127_agentn...
11.
Show HN: Aquifer – a control plane for agentic API traffic(github.com)
1 points by rjpruitt16 4mo ago|0 comments
12.
by rjpruitt16 5mo ago
Location: Memphis, TN (open to remote US) Remote: Yes Willing to relocate: Yes Technologies: Elixir (BEAM/OTP), Go, Python, Java, TypeScript, React, AWS (Lambda, SQS, DynamoDB, Kinesis), Kubernetes, Docker, distributed systems, o
13.
by rjpruitt16 5mo ago
I’m curious if you are looking into gleam at all. I have written elixir/gleam interop otp library. Probably will help a lot for safety in finance. I could help turn certain components into for type safe gleam code
14.
by rjpruitt16 5mo ago
Yeah, thanks for checking it out. I know it’s a lot read. Keep me in mind if you run retry storms for agents.
15.
Layer 8: The coordination protocol AI agents and embedded devices don't have yet(ezthrottle.network)
2 points by rjpruitt16 5mo ago|2 comments
16.
by rjpruitt16 6mo ago
Where do you think AI web agents with be by 2027? How are people creating value today? Where will browsers use be at the end of the year?
17.
by rjpruitt16 6mo ago
An api rate limit router - it’s gives a queue per user to solve noisy neighbors, it allows the user user to reroute 500s to other regions automatically, the api provider can tune the speed of request with headers. It should be a massive spe
18.
by rjpruitt16 6mo ago
Ezthrottle works by sending the request and depending on what error code the user wishes to reroute on, it will send to another region. It give the user a chance to say something different in case the api misclassifies the error. The user w
19.
by rjpruitt16 6mo ago
Cool stuff.
20.
by rjpruitt16 7mo ago
Thanks for this. I was in the camp of trust the LLM but y’all have made valid points. After discussing with ChatGPT, it agree there are some areas where it should not be trusted as accurate, but it said with historical facts like the holoca
21.
by rjpruitt16 7mo ago
What do you use?
22.
by rjpruitt16 7mo ago
Im curious if you guys are seeing rate limiting issues. Agents sharing api keys tend to be retry storm monsters. I wonder how agent companies will address
23.
Ask HN: How to scale agent systems when Layer 7 is unreliable?
1 points by rjpruitt16 7mo ago|2 comments
24.
by rjpruitt16 7mo ago
interesting. What type of features did this enable. How was it maintaining redis. How many queues did you have.
25.
Ask HN: How do you handle API rate limits in production?
3 points by rjpruitt16 7mo ago|2 comments
26.
Show HN: Stop Losing LangGraph Progress to 429 Errors(ezthrottle.network)
1 points by rjpruitt16 8mo ago|0 comments
27.
by rjpruitt16 8mo ago
Fair pushback — to clarify, I’m not assuming incompetence or suggesting infra should paper over bad architecture. By “losing sleep” I really mean on-call fatigue during partial outages — the class of incidents where backoff, shedding, and b
28.
by rjpruitt16 8mo ago
Agree backoff+jitter is table stakes, and load shedding/backpressure is necessary under sustained overload. The tricky cases I’m digging into are shared rate limits (429s) and many concurrent clients/agents where local backoff isn
29.
Ask HN: Is anyone losing sleep over retry storms or partial API outages?
2 points by rjpruitt16 8mo ago|4 comments
30.
Ask HN: Are retries the wrong abstraction under rate limits?
3 points by rjpruitt16 8mo ago|0 comments

More ›