What changed

This blog used to read like a broad notebook on machine learning systems. It is becoming something more focused.

The center is now AI Software Engineering: the discipline of building AI systems that are understandable, reproducible, observable, and production-ready once models meet infrastructure, interfaces, and real users.

Inside that center, four themes matter most here:

  • LLMOps as the evolution of MLOps
  • GCP as an operating context for production AI
  • Edge computing and agentic AI
  • Music analysis with NLP and LLMs

That is the map for everything else on the site.


Reading paths

1. LLMOps: what changes when the model becomes part of a larger system

The MLOps discipline that emerged around 2020 was designed for classical ML: feature pipelines, model registries, batch inference, drift detection on tabular data. LLMs broke most of those assumptions. This is where I work through what the new discipline looks like.

2. AI Software Engineering: the system around the model

Training a model is only one slice of the problem. The more durable work is in architecture, evaluation, testing, interfaces, cost, monitoring, and operational tradeoffs.

3. GCP, edge computing, and agentic systems

This is where the blog moves from abstract engineering language to concrete system constraints: hardware, latency, cost, deployment surface, and the problem of autonomy outside a single request-response loop.

4. Music, NLP, and LLMs

This is the parallel line of work that keeps the technical writing honest. Music is a strong test case because language models often look smarter than they are when metaphor, repetition, symbolism, and structure are doing the real work.

Full Music + NLP section


How to read the site now

If you want the clearest picture of the blog’s editorial direction, go next to AI Engineering.

If you want the research line that is least like everyone else’s AI writing, go to Music + NLP.

If you want updates when something genuinely new is ready, the best formats are RSS and the newsletter. I write when I have something worth saying, not to satisfy a content calendar.

GitHub · LinkedIn · X