Microservice Architecture & Performance
Navigating the trade-offs of distributed systems, latency bottlenecks, and high-performance backend design.
Production lessons, system design, and backend realities.
Deep dives into C++ interview questions, OS internals, memory management, and systems programming patterns.
Explore the architecture of a modern browser, multi-process models, and verify bugs in a massive C++ codebase.
Making LLMs run faster on CPU-only hardware through profiling-driven optimization of quantization, memory, and compute.
Designing distributed systems that don't just scale, but stay fast under operational pressure.
Navigating the trade-offs of distributed systems, latency bottlenecks, and high-performance backend design.
Phase 3: Deep dive into the technical challenges faced during the cross-compilation and emulation pipeline, and how they were resolved.
Why exponential backoff wasn’t enough, and how we learned that clients are part of the infrastructure.
A deep-dive into making Large Language Models run faster on CPU-only hardware through profiling-driven optimization of quantization, memory, threading, and compute.
Phase 1 results of the LL_LLM project: measuring latency, throughput, and memory usage of a 1.1B model on CPU.
Phase 2 of the cross-compilation series: Integrating the SEAL binary into Buildroot rootfs via overlays and executing it on an emulated Cortex-A7.
A comprehensive guide to building a reproducible environment for emulating ARM Cortex-A7, booting Linux via Buildroot, and cross-compiling Microsoft SEAL using Docker.
A comprehensive guide to implementing core data structures from scratch in C++. Essential for coding interviews.
A chronological tour of how C++ transformed from 'C with Classes' to a modern powerhouse.
From ACID to CAP, SQL to NoSQL - everything you need to know about choosing the right database.
A draft plan for contributing to the Gecko codebase.
A deep dive into building an async, text-based paper trading bot with Rust and Tokio.
Why another engineering blog? Because reliability is a feature, and trade-offs are everything.
How to scale a global tweet delivery system with 100M+ DAU.