Category: Technology
Batch to Streaming: What Actually Breaks (and When It’s Worth It) 

Batch to Streaming: What Actually Breaks (and When It’s Worth It) 

Right now, somewhere in a planning meeting, someone’s asking “can we make this real-time” and everyone’s nodding. Sounds much better, obviously. New data. Fast decisions. Dashboards on the move. Who would advocate for slower?  I’ve made this jump a few times in...

Listwise Knowledge Distillation for Dense Retrieval

Listwise Knowledge Distillation for Dense Retrieval

Modern search and RAG systems are powered by dense retrieval; however, practitioners are confronted with a frustrating tradeoff: bi-encoders are fast but imprecise, while cross-encoders are accurate but too sluggish for production. The operational complexity of...

Evidence-first Design For Consumer Health Formulations 

Evidence-first Design For Consumer Health Formulations 

Consumer health products are developed or formulated based on  marketing trends, trending “hero” ingredients, supplier stock formulations, and/or formulation expert subjectivity, and sometimes research. While these ways have  historically made some market...

When 2 stage model breaks causal inference 

When 2 stage model breaks causal inference 

A subtle selection bias in machine learning-based causal inference can flip your results — here’s how to spot and fix it  Machine learning is great at predicting outcomes. But when we try to use it to estimate causal impact, especially in a two-stage setup,...

What Building AI Chatbots Taught Me About Simplicity

What Building AI Chatbots Taught Me About Simplicity

I spent months overengineering an AI chatbot. Then I threw most of it away and got better results in two weeks. This is an anecdote of what actually worked when I built production RAG systems at scale, and why the lessons surprised me. The Complexity Trap I was tasked...