Sci Math && Code

Blog

Notes on software engineering, math and computer science.

01

Next.js for Kubernetes: one binary, one cache, and 2.4× more traffic on the same CPU

I compiled Next.js into a single Bun executable and gave it a Redis cache shared by every replica. Along the way I learned how to hunt silent failures, how to distrust my own benchmarks, and when to walk away from an idea that works.

software-architecture · nextjs · bun · kubernetes · redis

02

The Art of Destroying Software 🥋

The code you wrote today is probably wrong.

software-engineering · code-quality · refactoring

03

I published that SDD didn't work. Then I had to make it work.

From criticizing Spec-Driven Development to building an Agentic Governance Framework from scratch. What works, what doesn't, and what almost nobody is saying about the developer's role when AI is writing the code.

ai · software-engineering · agf · sdd · governance

04

Your CS Degree Is Working. You Just Can't See It Yet.

I spent years thinking theory and industry were two separate worlds. I was wrong. This is what a Computer Science degree quietly teaches you, whether you notice it or not.

cs · software-engineering · opinion

05

AI in software development: strategic decisions, and a few uncomfortable questions

"3x productivity with AI." Productivity measured as what? There is a huge difference between AI working faster and your company delivering more value. Questions that never make it into the 10-minute demo.

ai · software-engineering · opinion

06

What Happens When an LLM 'Thinks': Tokens, Logits, and Sampling

A deep dive into the three fundamental concepts that explain LLM behavior: tokens, logits, and sampling strategies.

ai · software-engineering · natural-language-processing · transformers · llm

07

The Patient Who Thought He Had Microservices (Spoiler: He Didn't)

A story about Distributed Monolith Syndrome and how Event-Driven Architecture is the cure.

software-architecture · microservices · event-driven · distributed-systems

08

From Raw C++ to 'Plug-and-Play' AI: Why the Human Spark Is Still the Real Engine

AutoML, LLMs and pre-built pipelines automate almost everything — yet the humans who understand the fundamentals are more valuable than ever.

ai · software-engineering · opinion

09

How to Build a RAG System Locally

Building a RAG System Locally using Ollama and PostgreSQL with PgVector

nlp · llm · rag · ollama