Blog by Parsa Alizadeh
Articles by Parsa Alizadeh on building for the web, mobile and AI: what worked, what broke, and why. Every article cites its sources.
14 articles shown
How language models read text: tokens, byte pair encoding, and why Persian costs more
What a token is, how byte pair encoding builds a vocabulary, why Persian and Turkish take more tokens than English, and how to count tokens instead of guessing.
AI4 min read
30,000 points, no 3D library: how the background of this site is drawn
How the particle field behind this site is drawn in raw WebGL: one draw call, five shapes blended in a shader, and the choices that keep it accessible.
Web4 min read
How to evaluate a RAG system: retrieval metrics, faithfulness and LLM judges
Measure retrieval and generation separately, with recall@k, MRR and nDCG for the retriever and faithfulness for the answer, and what research says about using language models as judges.
AI5 min read
Quantizing large language models: how 8-bit and 4-bit weights work
Why a 7-billion-parameter model needs 14 GB in 16-bit, how quantization shrinks it, why outliers make it hard, and what LLM.int8(), GPTQ, AWQ and QLoRA each changed.
AI5 min read
Why search engines and AI assistants mix up people with the same name
How entity linking decides which person a name refers to, why research finds language models weakest on less-known people, and what makes your own profiles easy to tell apart.
Web5 min read
Building a personal website that search engines and AI can understand
A first-hand walkthrough of how this site states who it is about, how it connects articles and projects to one identity, and what the published guidance says about each choice.
Web3 min read
What Core Web Vitals measure, and what fixing LCP on my own site taught me
The three Core Web Vitals and their thresholds, why lab scores are not the assessment, and a real Largest Contentful Paint investigation on this website.
Web4 min read
Generative Engine Optimization: what the research actually measured
A close reading of the 2023 paper that introduced Generative Engine Optimization, which methods improved visibility in AI-generated answers, which did not, and what the results do and do not show.
Web3 min read
React Native's New Architecture: what replaced the bridge, and why it matters
What the React Native New Architecture changes, JSI, Turbo Modules, the Fabric renderer and the event loop, based on the official documentation and the 0.76 release announcement.
Mobile3 min read
Motion that respects people: prefers-reduced-motion and WCAG
What WCAG says about moving content and interaction-triggered animation, how the prefers-reduced-motion media feature works, and the patterns this site uses to honour it.
Web3 min read
Retrieval-augmented generation: from the original paper to a working pipeline
What RAG is according to the 2020 paper that named it, the parts of a practical pipeline, and what research says about where to put retrieved text in the prompt.
AI3 min read
Structured data for a personal website: Person, ProfilePage and sameAs
How to describe yourself to search engines with JSON-LD, what Google's ProfilePage documentation asks for, and the rules that keep structured data honest.
Web3 min read
The Transformer, from the original paper: attention explained for developers
How scaled dot-product attention, multi-head attention and positional encoding work, with the formulas and settings taken directly from "Attention Is All You Need".
AI3 min read
How vector search finds neighbours fast: embeddings, similarity and HNSW
Why exact nearest-neighbour search does not scale, what approximate search trades away, and how the HNSW graph from Malkov and Yashunin's paper finds close vectors quickly.
AI3 min read