Performance

A visualization of neural network weights transitioning from 32-bit float to 8-bit integer representations.

Building a Production-Grade Model Quantizer: A CV Side Project

A hands-on guide to building a neural network model quantizer from scratch. This project bridges machine learning and systems engineering, signaling real production skills to hiring managers.

Short description of the cover image subject.

Deep Dive into CUDA Graphs for Efficient Kernel Launch Overhead Reduction

CUDA graphs reduce kernel launch overhead by batching operations. They can deliver up to 40% speedups in real-world inference pipelines.

Optimizing React Rendering with Intersection Observer: Lazy Loading at Scale

A practical guide to using Intersection Observer for component and image lazy-loading in React, with patterns for scaling to hundreds of list items, infinite scroll, and real-world performance metrics.

A diagram of a transformer with paged attention and LRU cache.

Building a Pure-Python Paged-Attention Transformer with LRU KV-Cache Eviction

Implement a paged-attention transformer with LRU KV-cache eviction in pure Python, demonstrating systems engineering proficiency.

A visualization of Varnish Cache purge flows and BAN table lookups

Optimizing Varnish Cache Purges with BAN Tables and Incremental Invalidation

Explore how Varnish Cache BAN tables and incremental invalidation strategies reduce purge overhead, keep cache coherent, and cut origin load in high-traffic production environments.