Napkin math

Napkin math
8 minutes by Simon Eskildsen

The goal of this project is to collect software, numbers, and techniques to quickly estimate the expected performance of systems from first-principles. For example, how quickly can you read 1 GB of memory? By composing these resources you should be able to answer interesting questions like: how much storage cost should you expect to pay for logging for an application with 100,000 RPS?

Free course series: How AI fits into real development workflows
sponsored by Jetbrains

AI is already part of everyday development, but using it effectively takes more than prompts and demos. This free, hands-on course series explores how developers code, test, review, and collaborate with AI in the context of real IDE workflows, using practical tasks and real projects throughout. Enroll for free.

Unconventional PostgreSQL optimizations
8minutes by Haki Benita

When it comes to database optimization, developers often reach for the same old tools: rewrite the query slightly differently, slap an index on a column, denormalize, analyze, vacuum, cluster, repeat. Conventional techniques are effective, but sometimes being creative can really pay off.

The challenges of soft delete
5 minutes by atlas9

Software projects often implement "soft delete", maybe with a deleted boolean or an archived_at timestamp column. If customers accidentally delete their data, they can recover it, which makes work easier for customer support teams. Perhaps archived records are even required for compliance or audit reasons.

Getting real with LLMs
10 minutes by Gilad Peleg

Gilad breaks down real-world coding tasks into four categories to show where AI tools actually help versus create hype. Current AI excels at simple tasks with no side effects and complex isolated projects. It struggles with simple changes that could break other systems and complex multi-system tasks. Most production code falls into these problematic categories where AI adds risk rather than value.

Forecasting models to improve driver availability at airports
15 minutes by Bob Zheng

Uber developed three AI models to improve airport rideshare operations, which make up 15% of their global bookings. The models predict driver wait times, hourly earnings potential, and driver shortages at airports. These predictions help drivers decide when to work at airports and allow Uber to summon drivers during busy periods. The system reduced driver wait uncertainty and improved rider pickup times.

And the most popular article from the last issue was:

• I got paid minimum wage to solve an impossible problem