GGeekyAnts4d ago · 15 min readTaming Elasticsearch at ScaleElasticsearch is useful for searching an increasing dataset that starts adding pressure on the database and the application. Instead of checking millions of records one by one, it builds an index thattechblog.geekyants.com
AAsep SayyadOct 20 · 19 min readThe Mystery of the Dying 1GB ServerHow an unsuspected GUI daemon, zero swap, and unrotated Docker logs pushed our cloud node into kernel panics (and how we reclaimed 500MB of free RAM). The incident report pinged my phone at 02:40 AM. asepsayyad007.hashnode.dev
Sspipaliya8Oct 12 · 6 min readEF Core Any vs Count: Which One Is Better for Existence Checks?When working with Entity Framework Core (EF Core), we often need to check whether a record exists in the database. For example: Does a student already exist? Has an email address already been registsudhirpipaliya.hashnode.dev
MMichael LohrOct 10 · 13 min read Smallest "Hello World" continuedNearly two years ago, I wrote about the smallest possible "Hello World!" program. It's one of my most read blog posts. Recently, I started playing around a lot with AI (like almost everybody else), soblog.lohr.dev
AAhmed AdawyOct 4 · 5 min readDemystifying LLM Serving Infrastructure: How PagedAttention and Continuous Batching Scale InferenceMoving a Large Language Model (LLM) from a local prototype in a Jupyter notebook to a high-throughput, multi-tenant production environment is a brutal awakening. While data scientists spend months optadawy.hashnode.dev
AAhmed AdawyOct 2 · 3 min readLLM Inference Engineering: Overcoming the KV-Cache Bottleneck and Maximizing Production ThroughputLLM Inference Engineering: Overcoming the KV-Cache Bottleneck and Maximizing Production Throughput When moving a Large Language Model (LLM) from a prototype to a production environment, you immediateladawy.hashnode.dev
MMikhail PanfilovSep 29 · 22 min readCan safe Rust ever beat Google's C Brotli?Brotli, SIMD, bounds checks, and what it takes to compete with Google's C implementation. At one point, mbrotli's greedy match finder had the right AVX2 token and no native vector compares. The generamnwa.hashnode.dev
GGonzalo AguilarSep 28 · 12 min readWhen Personalisation Costs 15 Seconds: Redesigning an LLM Feature for ReliabilityHow I changed personalised vocabulary practice so learners could start a session without waiting for generated sentences. Parlato already had a usable exercise sentence, yet loading a learning session22EBgonzalosimon.hashnode.dev
Ggayatrikakumanu25Sep 24 · 17 min readFigma-to-Code at Scale: What Actually Drives Cost, Quota, and QualityOr: why the "smart" way to fetch Figma designs turned out to be the expensive way. TL;DR: The "fetch a lightweight outline first" advice is backwards — it cost almost as much as fetching everything gayatrikakumanu.hashnode.dev
GGulshan YadavSep 22 · 12 min readReducing Model Size Without Losing Accuracy: QuantizationA practical deep dive into model quantization — the precision-ladder trick that takes a 14GB model down to 3.5GB, what you lose, what you keep, and the code to measure both. Six months ago I was tryinmrgulshanyadav.hashnode.dev
