Featured
One Scheduler, Many Workers: Running Luigi Across Multiple Servers with Zero Downtime
Growing a single-VM Luigi pipeline into a fleet: a highly available luigid scheduler, multiple workers, and zero-downtime GitLab CI/CD.
01 Aug 2026
LeetCode 1422: Maximum Score After Splitting a String
A prefix-sum solution to LeetCode 1422, Maximum Score After Splitting a String, evaluating every split in O(n) time.
01 Jan 2025
All Stories
LongLLMLingua: Prompt Compression for Long-Context LLMs
LongLLMLingua's question-aware compression, document reordering and subsequence recovery, for prompts that run to long contexts.
09 Sep 2024 · 3 min read
LLMLingua: Compressing Prompts for Accelerated LLM Inference
How LLMLingua compresses prompts using a budget controller, iterative token-level compression and distribution alignment.
01 Sep 2024 · 5 min read
LLaMA: Open and Efficient Foundation Language Models
A breakdown of Meta's LLaMA: the transformer modifications, the training corpus, and the training procedure behind the model.
01 Dec 2023 · 3 min read
The YOLO Algorithm, End to End
Every piece of this series assembled into one detector: the grid, the output tensor, the box parameterisation that makes training stable, and the full pipeline.
03 Jun 2023 · 5 min read