DEV Community

#llm

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
We Put Our Router on an Academic Benchmark. Here Are the Numbers We'd Rather Hide.

We Put Our Router on an Academic Benchmark. Here Are the Numbers We'd Rather Hide.

1
Comments 1
5 min read
My routing policy and my traces disagreed 96 times. Never once on the main thread.

My routing policy and my traces disagreed 96 times. Never once on the main thread.

1
Comments 4
14 min read
Securing Agentic AI for Singapore Enterprises: A Reference Architecture

Securing Agentic AI for Singapore Enterprises: A Reference Architecture

Comments
3 min read
Building an Observable AI Market Research Agent with SigNoz

Building an Observable AI Market Research Agent with SigNoz

Comments
1 min read
Loop Engineering Is Mostly Papering Over a Model That Won't Converge

Loop Engineering Is Mostly Papering Over a Model That Won't Converge

2
Comments 2
4 min read
I Trained a 6.4M-Parameter Transformer From Scratch to Talk About Recipes

I Trained a 6.4M-Parameter Transformer From Scratch to Talk About Recipes

1
Comments
5 min read
Empire LLM for Codex: AI Code Review Without the Chaos

Empire LLM for Codex: AI Code Review Without the Chaos

Comments
5 min read
Is Speculative Decoding's Speedup a Hardware Problem or a Model Problem?

Is Speculative Decoding's Speedup a Hardware Problem or a Model Problem?

Comments
8 min read
I built a CLI that tells you if your codebase fits an LLM's context window

I built a CLI that tells you if your codebase fits an LLM's context window

4
Comments
2 min read
I built a production AI agent as a Honda service advisor. Then I read the textbook.

I built a production AI agent as a Honda service advisor. Then I read the textbook.

Comments
7 min read
How Do You Contain an AI Agent Failure You Can't Prevent?

How Do You Contain an AI Agent Failure You Can't Prevent?

1
Comments
2 min read
CacheGuard

CacheGuard

Comments
6 min read
Claude Opus 5 leads on agentic work — and undercuts Fable 5 on cost

Claude Opus 5 leads on agentic work — and undercuts Fable 5 on cost

Comments
2 min read
OpenAI's model escaped its sandbox and hacked Hugging Face to cheat on a test

OpenAI's model escaped its sandbox and hacked Hugging Face to cheat on a test

Comments
3 min read
Stress-testing my Multi-LLM engine: 93 chunks, 8 models, and one "Insufficient Balance" error.

Stress-testing my Multi-LLM engine: 93 chunks, 8 models, and one "Insufficient Balance" error.

Comments
2 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.