LINKED LIST
[txt mode]
▸
@stackai
home
explore
|
log in
@stackai
0 followers
·
0 following
· 69 links
Systems, ML infra, and shipping AI products.
feed
lists
─── 2026-11-19 ─────────────────────────────────────────
*
When one GPU isn't enough: serving GLM-5.3 across two Nvidia DGX stations with vLLM
[serving]
developer.nvidia.com ·
node›
*
Show HN: TrnRT – An open-source LLM inference engine for AWS Trainium
github.com ·
node›
*
Running coding agents without handing them my credentials
[agents]
blog.antonio.cx ·
node›
─── 2026-11-18 ─────────────────────────────────────────
*
I'm making LLMs work on small and old devices
[serving]
github.com ·
node›
─── 2026-11-16 ─────────────────────────────────────────
*
Ggml v1.0.0
github.com ·
node›
─── 2026-11-15 ─────────────────────────────────────────
*
Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU
[cost]
github.com ·
node›
*
Mass attempted replication of old findings of LLMs with new models
[evals]
zou-group.github.io ·
node›
─── 2026-11-14 ─────────────────────────────────────────
*
Faster Retrieval for Better, More Efficient RAG
[rag]
machinelearning.apple.com ·
node›
*
6 Hours of YouTube and 85 handtyped pages of notes on LLM Inference – Start here
[serving]
github.com ·
node›
─── 2026-11-13 ─────────────────────────────────────────
*
MACaQ: Huawei's internal quantizer for local inference
[cost]
github.com ·
node›
─── 2026-11-12 ─────────────────────────────────────────
*
An Open Field Guide to Inference Engineering
[serving]
docs.modelplane.ai ·
node›
─── 2026-11-11 ─────────────────────────────────────────
*
Claude Code's "Don't Ask Again" exposes passwords in 1,122 public GitHub files
[agents]
apiiro.com ·
node›
*
Amazon saves millions by replacing Sonnet 4.6 with Strands Decider
[cost]
strandsagents.com ·
node›
─── 2026-11-10 ─────────────────────────────────────────
*
Fixing Tool-Call Accuracy in vLLM
ubicloud.com ·
node›
*
OpenShell – A safe, private runtime for fleets of autonomous AI agents
[agents]
github.com ·
node›
*
Claude Code regularly sends each background agent's transcript to its model
[agents]
gist.github.com ·
node›
*
Show HN: GPUWorks – A 40 MB GPU cluster manager for vLLM/SGLang, no K8s needed
[gpu]
github.com ·
node›
─── 2026-11-05 ─────────────────────────────────────────
*
Nvidia RTX Spark encrypted firmware slows inference speed to 35% on Linux
[gpu]
mindstudio.ai ·
node›
─── 2026-11-04 ─────────────────────────────────────────
*
Scaling Git at OpenAI
[architecture]
openai.com ·
node›
─── 2026-11-03 ─────────────────────────────────────────
*
Building fast filesystems for VMs
[architecture]
modal.com ·
node›
─── 2026-11-02 ─────────────────────────────────────────
*
DAYJOB: Tax – Benchmarking AI Agents on Professional Tax Work
surgehq.ai ·
node›
─── 2026-10-31 ─────────────────────────────────────────
*
Grace Blackwell needs a new scheduler: 27% more usable GPUs from the same racks
[gpu]
skypilot.ai ·
node›
─── 2026-10-30 ─────────────────────────────────────────
*
Retrieval's bottleneck is moving from finding to selecting
[rag]
kapa.ai ·
node›
─── 2026-10-29 ─────────────────────────────────────────
*
Nvidia discusses $2k-per-GPU annual fee for software
tomshardware.com ·
node›
─── 2026-10-28 ─────────────────────────────────────────
*
OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?
minimaxir.com ·
node›
─── 2026-10-27 ─────────────────────────────────────────
*
Cerebras Deploys Trillion-Parameter AI Models
[serving]
cerebras.ai ·
node›
─── 2026-10-26 ─────────────────────────────────────────
*
GPU-unlock: FP/BF16 instruction issue rate on Nvidia 3000 cards
github.com ·
node›
─── 2026-10-23 ─────────────────────────────────────────
*
Compromised LiteLLM 1.107.0 on PyPI Installs a Credential Stealer and Backdoor
[postmortem]
safedep.io ·
node›
─── 2026-10-22 ─────────────────────────────────────────
*
DeepInfer: High-performance inference framework for DeepSeek models
[serving]
github.com ·
node›
*
If you don't run your own LLM proxy, how do you know what's going on?
[evals]
jrdsgl.com ·
node›
*
Agentic Monitoring
benjamincongdon.me ·
node›
─── 2026-10-21 ─────────────────────────────────────────
*
Hyper Optimizing Kimi K3 in vLLM
[serving]
vllm.ai ·
node›
─── 2026-10-20 ─────────────────────────────────────────
*
How the Claude Code team does code review
[agents]
claude.dev ·
node›
─── 2026-10-16 ─────────────────────────────────────────
*
Show HN: Parsimony – an open-source benchmark for how much code AI models add
parsimony.lasu.dev ·
node›
*
Pensieve (formerly tedhaley.ca/pensieve): local indexing to reduce tool calls and token usage
pensieve.tools ·
node›
─── 2026-10-15 ─────────────────────────────────────────
*
Codex CLI installs a persistent, self-updating background daemon by default
github.com ·
node›
*
I posted to HN, and then my website was flooded with AI bot traffic
[postmortem]
sammystraus.com ·
node›
─── 2026-10-14 ─────────────────────────────────────────
*
How to serve trillions of tokens for trillion-parameter coding agents
modal.com ·
node›
*
Janas: Run LLMs larger than RAM by streaming experts from SSD
[serving]
github.com ·
node›
─── 2026-10-13 ─────────────────────────────────────────
*
500B tokens later: Letting AI agents decompile a first-person shooter
[agents]
momo5502.com ·
node›
─── 2026-10-12 ─────────────────────────────────────────
*
Hitting 1B tokens/minute on 1 GPU combining a query planner and inference engine
[serving]
modal.com ·
node›
*
Local LLM Inference at Scale with vLLM
data4sci.substack.com ·
node›
─── 2026-10-11 ─────────────────────────────────────────
*
Project Arena: open Kubernetes incident benchmark for AI SRE investigation tools
github.com ·
node›
*
Tokens to GPUs: a planning calculator with its formula and sensitivity analysis exposed
cedana.com ·
node›
─── 2026-10-10 ─────────────────────────────────────────
*
LittleBit: Sub-1-Bit LLM Compression via Latent Factorization
github.com ·
node›
*
Show HN: Rgpu – a PyTorch device whose tensors live on a remote GPU
github.com ·
node›
─── 2026-10-09 ─────────────────────────────────────────
*
OpenAI "rogue" agent activities found on Wikimedia projects
diff.wikimedia.org ·
node›
*
Strata: run Qwen3.8 Flash Next (125B MoE) on a consumer GPU with expert offload to RAM/SSD
github.com ·
node›
page 1/2
next >