Erturk
BlogBook a call

Mehmet
Erturk

EngineerDesignerFounderBuilder who ships.

Not tools. Not dashboards. Systems that run work autonomously.
9+ years in commerce, payments, and AI agents — zero to scale with P&L ownership. Hands in the code.

Recent

All 65 articles →

From Hard to Harnessed: Making On-Device LLM Tool-Calling Actually Work→

Jun 3, 2026

A 3B model on your phone should log "5$ starbucks" as a five-dollar transaction attributed to Starbucks. Mine wrote a coffee-shop review instead. Here is the harness that fixed it.

User-Configurable Cache TTL: A Pattern Stolen from Magento for Multi-Tenant SaaS→

Apr 5, 2026

How I gave each workspace control over its own cache lifetime, with auto-flush on TTL decrease, cross-entity invalidation, and E2E tests that prove it works.

Practical Allocation Hunting in Rust: Patterns That Actually Matter→

Apr 3, 2026

Real before-and-after patterns for eliminating unnecessary heap allocations in production Rust, from ByteCounter to LazyLock.

Query-Time Media Resolution in SurrealDB: Pushing Transforms into DEFINE FUNCTION→

Apr 1, 2026

Push media URL resolution into SurrealDB functions instead of post-processing API responses in your application layer.

Managing Concurrency in AI Workflows: Parallel Branches and Shared State→

Mar 15, 2026

How we handle concurrent workflow execution in Rust using tokio::spawn and RwLock, the limitations we've hit, and where we're going next.

GitHubLinkedInXYouTubeRSS
© 2026