Large Language Models (LLMs) are only as good as the data that shapes them. Whether you’re fine-tuning a model for domain-specific use or building an LLM-powered application from scratch, data readiness is the single biggest factor influencing performance, reliability, and ethical outcomes. As the saying goes, “garbage in, garbage out”  - but with LLMs, the cost of poor data goes beyond technical glitches; it can lead to bias, misinformation, and loss of trust. In this post,

In the previous blog post,  we explored how LLMs are changing the way we develop software and how we need to think differently about data, engineering, and architecture to cater to them. In this second part, I will explore testing, observability, and the related aspects of ethics a little further. And when it comes to LLM-powered applications, testing is certainly not second fiddle. In fact, it could be argued that it is even more important than in traditional software develo

While I’ve written on AI many times, I do often resist writing on the topic because I fear it is often overplayed and somewhat of a potential tech bubble. It won’t be going away, though, and even if its impact wanes, we still need to learn how to test and build quality around it. Especially in the realm of LLMs, which have had the largest impact on software development, both in how we create software and the types of solutions we can solve. Large language models (LLMs) like G

I’ve spent the past two blog posts talking about the importance of change communication , particularly the transparency required  to...