Latest Insights
Exploring the frontiers of AI, software architecture, and engineering.
Prepared Statements in PHP and PDO: Killing SQL Injection by Construction
Every SQL injection in history came from gluing strings into queries. Prepared statements make the category impossible — if you use them everywhere, including LIMIT.
Read Article →p50, p95, p99: Percentiles for Monitoring, SLAs and Honest Latency Talk
Average latency is 120ms — and 5% of users wait over 3 seconds. Averages hide the suffering tail; percentiles price it. How to compute, plot, and SLA them.
Read Article →Finding the Rhythm: Seasonality Detection in Business Data
Sales dip every Sunday. Traffic peaks on Wednesdays. If you do not model the rhythm, your forecasts and anomaly alerts will keep crying wolf — here is how to detect cycles properly.
Read Article →What a Confidence Interval Actually Means in Forecasting
A forecast without error bars is a guess wearing a suit. But most forecast bands promise more than they deliver — here is what a 95% interval actually means and how to keep it honest.
Read Article →Grading Your Data: A Practical Data-Quality Scorecard
A data-quality grade turns vague unease about your spreadsheet into a number you can act on. Here is the measurement framework: completeness, uniqueness and validity, scored per column.
Read Article →Beyond Counting: What Categorical Columns Actually Say
Counts of categories hide the interesting part: concentration. A Herfindahl index and a top-share number reveal whether one category dominates or everything is fragmented.
Read Article →