# samir-sarwar/samir-vllm

> An educational LLM inference engine in C++ and CUDA that walks through what happens between a prompt and a generated token, with a video series and handwritten notes.

- Magnitude: 6.4 out of 10 — Early signal
- Stars: 167 total · +13 stars measured 2026-11-10, 00:26–06:03 UTC, ≈ 109 by evening
- Star trust: star growth looks organic
- Category: Learning & lists · Language: C++ · Created: 2026-08-27 · Last push: 2026-11-03
- GitHub: https://github.com/samir-sarwar/samir-vllm · Page: https://gitnova.dev/en/r/samir-sarwar/samir-vllm

## Useful for

- Walk through how prefill, decode and the KV cache work in a transformer
- Study a C++/CUDA implementation of paged attention and continuous batching
- Run generation on Llama 3.2 1B Instruct and inspect page pool stats

## Why it’s here

- Star-counter measurements on 2026-11-10 (UTC), 00:26–06:03: 154 → 167 stars (+13). This is the change over that interval.
- Estimated end-of-day forecast: about +109 stars, using observed gains and the previous day.
- Before this, the repository barely got any stars — about 1 per day.
- Hacker News: “6 Hours of YouTube and 85 handtyped pages of notes on LLM Inference – Start here” — 77 points, 18 h ago.
- About 7 forks a day — people are taking the code.
- Recent forks include notable developers: @toranb (370 followers).

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 6
- Issues and pull requests: 0
- Watchers: 0
- Average over the last week: 40 per day
- Usual pace: 0 per day
- Stars in the last hour (measured): 4

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-10-12 … 2026-11-10: 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 5, 1, 3, 0, 0, 158, 13

## Magnitude by day, last 2 days

11-09 5.4, 11-10 6.4

## Hacker News

- 6 Hours of YouTube and 85 handtyped pages of notes on LLM Inference – Start here — 77 points, 12 comments: https://news.ycombinator.com/item?id=50424199

## Spotted in now

- Spotted on Hacker News

## More in this category

1. **amosblomqvist/learnharness** — 5.5 · Breakout · Learning & lists · TypeScript · +169 stars measured 2026-11-10, 00:26–06:03 UTC, ≈ 905 by evening
   A desktop app that uses AI agents to plan and teach lessons from your current understanding to a goal, tracking progress in a persistent knowledge graph.
   Full card: https://gitnova.dev/en/r/amosblomqvist/learnharness.md
2. **awesome-dsh-plugin/awesome-dsh-plugin** — 4.4 · Steady · Learning & lists · JavaScript · +48 stars measured 2026-11-10, 00:36–06:03 UTC, ≈ 189 by evening
   A curated list of plugins for DeepSeek Harness (dsh) — a collection of extensions for this environment. Useful for those looking for ready-made dsh plugins.
   Full card: https://gitnova.dev/en/r/awesome-dsh-plugin/awesome-dsh-plugin.md
3. **public-apis/public-apis** — 4.0 · Steady · Learning & lists · Python · +92 stars measured 2026-11-10, 00:27–06:04 UTC, ≈ 352 by evening
   A community-curated list of free public APIs across dozens of categories, from animals and anime to finance, geocoding and weather. Aimed at developers looking for a ready-made API for a prototype or product.
   Full card: https://gitnova.dev/en/r/public-apis/public-apis.md
4. **DeanChensj/jiangshan-map** — 3.8 · Steady · Learning & lists · JavaScript · +100 stars measured 2026-11-10, 00:27–06:04 UTC, ≈ 313 by evening
   An interactive historical atlas of China from 403 BCE to the present, with 18 dynasties, 164 territory snapshots, place-name genealogy, and poetry routes, designed for history and classical poetry enthusiasts.
   Full card: https://gitnova.dev/en/r/DeanChensj/jiangshan-map.md
5. **rohitg00/ai-engineering-from-scratch** — 3.7 · Steady · Learning & lists · Python · +31 stars measured 2026-11-10, 00:27–06:04 UTC, ≈ 218 by evening
   Open-source AI engineering curriculum: 523 lessons across 20 phases (~342 hours) in Python, TypeScript, Rust and Julia, where every lesson ships a reusable artifact such as a prompt, skill, agent or MCP server.
   Full card: https://gitnova.dev/en/r/rohitg00/ai-engineering-from-scratch.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-11-10 06:13 UTC, updated every 30 minutes.