Welcome to My Blog Hi, I’m Allan. I’m an independent machine learning engineer and systems architect, working at the intersection of AI, high-performanc...
Local AI System: Sovereign ML Stack (Personal Architecture) I'll be documenting my local AI setup as a ongoing process but here's what I have now What I’ve built is not just a “local AI setup” in the typical sense. It is a tightly integrated system where local compute, cloud execution, and storage are deliberately separated but deeply interconnected , so that I can move from idea to large-scale…
If you're like me, you're probably always trying to squeeze as much performance out of your systems and make each GPU hour count. It's a constant battle against bottlenecks, inefficiencies, and the sheer complexity of modern hardware. For a while, I've been working on a project called PyC , a toolchain of technologies with some experimental methods I'm trying to run and see if they have meaningful…
Notes from Building and Operating NexaCompute One thing I didn’t fully appreciate early in my ML career is how much infrastructure shapes thinking . Not in an abstract sense, but very concretely: what experiments you try, how often you iterate, how confident you are in results, and how much mental energy you burn just keeping things alive. A lot of ML today works because engineers are smart and…
Technical Evaluation of NexaSci-Falcon-10B A Lightweight Scientific Reasoning Model This project began as a controlled experiment: take a curated 100k scientific dataset, distill it into a 10B parameter model, add a small agentic post-training layer, and evaluate it directly against frontier systems under identical conditions. The goal was not scale. The goal was to understand what a carefully…
End-to-end distillation, deployment, and inference R&D on dual RTX 4090s 10 Nov, 2025 TL;DR: In ~two weeks I stood up my training/inference platform ( Nexa_Compute ), distilled a 10B scientific model from ~100k GPT-4 Q&A pairs for about $200 , and then benchmarked it on a dual-RTX 4090 box using vLLM . The best bf16 config (no quant) reached ~878 tokens/sec with ~0.29 s average latency. The…
I never set out to design an engineering philosophy. There was no whiteboard session, no manifesto-writing moment, and no grand declaration of how I’d build things. What happened was simpler: I spent years building systems alone, under tight constraints, and a pattern quietly emerged. At first it was just survival—no racks of GPUs, no institutional safety net, no budget to burn. Just me, an idea,…
Introduction Protein secondary structure prediction is a cornerstone of bioinformatics. Understanding how proteins fold and interact helps researchers unravel biological function, design drugs, and engineer novel proteins. Traditional experimental methods like X-ray crystallography and NMR spectroscopy are highly accurate but slow and expensive. Machine learning (ML) offers a faster, scalable…
When it comes to software, I’ve always approached it differently than most. Not because I want to be contrarian, but because I wanted control, speed, and efficiency. Over the years, I’ve built a personal ecosystem of software that serves my needs first — and it’s taught me more about development, infrastructure, and creative problem-solving than any framework or methodology I’ve studied in…
Systems Engineering: Why Integration Matters More Than Specialisation The “specialist versus generalist” debate is everywhere in engineering circles. Some argue that deep focus wins; others say breadth is the future. I think both sides are missing the point. The real edge today lies in integration Systems engineers aren’t specialists or generalists; they’re integrationists The Burger Stack Analogy…