Jan 27 · 1 min read · Microsoft Foundry Evaluations provide a structured, repeatable way to measure the quality, safety, and reliability of generative AI systems—long after the first model is deployed. Instead of treating evaluation as a one‑time check, Foundry brings con...
Join discussion
Dec 30, 2025 · 4 min read · If you’re building AI agents and need production-grade open-source observability and evaluation without vendor lock-in, Arize Phoenix is worth exploring. Phoenix is an open-source AI development platform built on OpenTelemetry and OpenInference instr...
Join discussion
Dec 11, 2025 · 3 min read · Hallucinations are one of the biggest challenges in production AI agents. Here's how to detect them with Python: Five Types of Hallucinations in AI Agents Context Hallucinations: The agent invents facts not in the provided context. Tool Hallucinati...
Join discussion
Dec 10, 2025 · 4 min read · Building AI agents is hard. Evaluating them is harder. Most teams I talk to are evaluating their agents the wrong way. They look at the final output and ask, "Is it correct?" But that's like grading a math test by only looking at the final answer, no...
Join discussion
Nov 24, 2025 · 15 min read · The history of natural language processing evaluation reveals a persistent pattern. Metrics created for one generation of technology become dangerously inadequate for the next. When statistical machine translation systems dominated the field, BLEU sc...
Join discussion
Nov 7, 2025 · 12 min read · When you start building with LLMs, it quickly becomes clear that not all models behave the same. One model may excel at creative writing but struggle with technical precision. Another might be thoughtful yet verbose. A third could be fast and efficie...
Join discussion
Oct 20, 2025 · 6 min read · Course Lead Scoring Project This blog post details the process and findings of a machine learning project aimed at predicting whether a potential lead for an online course will convert (enroll). Note: the project link is here 1. Data Loading and Prep...
Join discussionJul 29, 2025 · 7 min read · Written by Ian Unsworth at MindsDB AI evaluations ("evals") have emerged as the critical success factor distinguishing enterprises that successfully deploy AI solutions from those trapped in pilot purgatory. Only 10% of enterprises have Gen AI in pro...
Join discussion
Jul 28, 2025 · 10 min read · Introduction We recently built a cell-typer model based on a subset of the CellXGene V2 dataset. This model is a stepping stone towards a “universal” cell-typer model that will allow us to annotate cell types (and other properties) based on single ce...
Join discussion
Jul 24, 2025 · 14 min read · Introduction I have been tinkering with LLMs at work and outside now for quite a while and one of the most pressing issues compared to traditional machine learning is the unsolved problem of how to evaluate them. Evaluating LLM outputs is exponential...
SSebastian commented