Blog

An AI agent harness is the software around an LLM that turns it into an agent. Learn how it works, see diagrams, and explore open-source harnesses.

Learn Jev in Python with Choice, Noul, and Score examples. Route AI agents, batch decisions, and compare costs and latency with OpenAI GPT-5.4 nano.

Explore iPhone launch prices adjusted for inflation with Python. Compare model lines, screen sizes, and real price trends in an interactive Mercury dashboard.

I analyzed 88,975 Hacker News job posts from 2012 to 2026. The board shrank, remote work peaked, pay became clearer, and senior roles took over.

789 GitHub incidents from 2022 to 2026. Reported incidents held flat for three years, then more than doubled. The longest incidents got much longer.

Explore GitHub incident history from 2022–2026 in a GitHub-style activity calendar, then build your own interactive outage dashboard with Python and Mercury.

Meet three new Mercury widgets for interactive Python data apps: ActivityCalendar, Sankey and Funnel, demonstrated with real GitHub and Hacker News data.

XGBoost multi-output regression explained: multi_output_tree vs one_output_per_tree, working code, and a benchmark on 2.7M rows of market data.

A practical, no-hype guide to running an LLM on your own machine — how to pick a model and a runtime, what hardware you need, and what local LLMs are actually good for, from coding to private data analysis.

Learn how to train a Machine Learning model with MLJAR AutoML, automatically generate a Web App, and publish it online for single and batch predictions.