Christopher Maher

Husband, dad, and software engineering leader. Passionate about automation, AI, emerging tech, and ham radio (N7CPM).

Joined on Mar 17, 2026

Christopher Maher

Jun 14

Making a fleet of self-hosted LLM agents trustworthy

#ai #llm #kubernetes #opensource

6 min read

Christopher Maher

Apr 29

TurboQuant on a MacBook Pro, part 2: perplexity, KL divergence, and asymmetric K/V on M5 Max

#ai #llm #kubernetes #opensource

8 min read

Christopher Maher

Apr 28

TurboQuant on a MacBook Pro: two findings the upstream discussion missed

#ai #llm #kubernetes #opensource

7 min read

Christopher Maher

Apr 27

62.2% on Aider Polyglot from a MacBook Pro. Then the other model we tried scored 4%. Here's what actually happened, with a working cost loop attached.

#kubernetes #ai #llm #opensource

16 min read

Christopher Maher

Apr 24

We ran Qwen3.6-27B on $800 of consumer GPUs, day one: llama.cpp vs vLLM

#kubernetes #ai #llm #opensource

15 min read

Christopher Maher

Apr 8

LLMKube Now Deploys Any Inference Engine, Not Just llama.cpp

#llm #opensource #ai #kubernetes

3 min read

Christopher Maher

Apr 6

I tested speculative decoding on my home GPU cluster. Here's why it didn't help.

#llm #kubernetes #gpu #ai

5 min read

Christopher Maher

Apr 3

Google Released Gemma 4 Yesterday. I Had It Fixing Real Bugs by Lunch.

#kubernetes #llm #homelab #ai

5 min read

Christopher Maher

Mar 30

I Tested TurboQuant KV Cache Compression on Consumer GPUs. Here's What Actually Happened.

#llm #kubernetes #gpu #ai

6 min read

Christopher Maher

Mar 23

The $0 Problem: Why Every Tool Says Your On-Prem Inference is Free

#finops #ai #kubernetes #opensource

4 min read

Christopher Maher

Mar 17

llama.cpp on Kubernetes: The Guide I Wish Existed

#kubernetes #ai #opensource #devops

9 min read

DEV Community

Christopher Maher

Badges

Writing Debut

Making a fleet of self-hosted LLM agents trustworthy

TurboQuant on a MacBook Pro, part 2: perplexity, KL divergence, and asymmetric K/V on M5 Max

TurboQuant on a MacBook Pro: two findings the upstream discussion missed

62.2% on Aider Polyglot from a MacBook Pro. Then the other model we tried scored 4%. Here's what actually happened, with a working cost loop attached.

We ran Qwen3.6-27B on $800 of consumer GPUs, day one: llama.cpp vs vLLM

LLMKube Now Deploys Any Inference Engine, Not Just llama.cpp

I tested speculative decoding on my home GPU cluster. Here's why it didn't help.

Google Released Gemma 4 Yesterday. I Had It Fixing Real Bugs by Lunch.

I Tested TurboQuant KV Cache Compression on Consumer GPUs. Here's What Actually Happened.

The $0 Problem: Why Every Tool Says Your On-Prem Inference is Free

llama.cpp on Kubernetes: The Guide I Wish Existed