Hey, I build platforms, lakehouse, streaming, data analytics.

Contributions to the Polars query engine (Rust), mostly optimizer, IO, and reliability:

• #28017 feat: Optimize joins with redundant equi-join keys
• #27727 / #28002 feat: Optimize len and null_count comparisons
• #27590 feat: Add Rust backend for Expr.has_nulls
• #28157 fix: Avoid slice pushdown panic on shared cache inputs
• #28225 fix: Classify S3 region request failures as IO errors
• #28066 fix(python): Close ResourceWarning leaks in database tests

All merged PRs: author:0guban0v is:merged · reported issues: author:0guban0v

• bayesian-llm-judge-reliability: Bayesian IRT analysis of LLM-as-judge reliability on JudgeBench. Main finding: judge rankings are not stable across item populations.