Benchmarking SlateDB vs. RocksDBnixiesearch.substack.com3 points·by shutty·w zeszłym miesiącu·0 comments
Show HN: MurrDB: A RocksDB-based NVMe/S3 cache for AI inference workloads1 points·by shutty·2 miesiące temu·0 comments
Lies, damned lies, and Elastic's benchmarksgouthamve.dev4 points·by shutty·2 miesiące temu·0 comments
Fine-tuning Qwen3 at home to respond to any prompt with a dad jokemedium.com1 points·by shutty·7 miesięcy temu·0 comments
The Day Our Own Queries DoS'ed Us: Inside Zalando Searchengineering.zalando.com1 points·by shutty·7 miesięcy temu·0 comments
Show HN: Fine-tuning Qwen3 at home to respond to any prompt with a dad jokenixiesearch.substack.com10 points·by shutty·7 miesięcy temu·0 comments
Show HN: I benchmarked read latency of AWS S3, S3Express, EBS and Instance storenixiesearch.substack.com3 points·by shutty·7 miesięcy temu·0 comments
I put a real search engine into a Lambda, so you only pay when you searchnixiesearch.substack.com48 points·by shutty·8 miesięcy temu·13 comments
We found embedding indexing bottleneck in the least expected place: JSON parsingnixiesearch.substack.com2 points·by shutty·8 miesięcy temu·0 comments
shutty·8 miesięcy temu·discussSurprisingly, 3GB is a real practical RAM limit for aws lambdas in 2025: you can only have more than that if you submit a support ticket. But it's not really mentioned anywhere in the docs.