~0x64616E69656C
personal engineering lab

Systems.
Security.
Odd machines.

I build, simplify and occasionally resurrect systems around e-commerce, search, web infrastructure and applied cryptography — plus one increasingly over-engineered Audi A2 diagnostic platform.

static HTML · no framework · no tracking Debian · Clang · C/C++ · PHP · Python · JS
01 / SELECTED WORK

Things that refuse to stay small.

Production systems, low-level protocol work, security lessons from real operations and one car project that somehow turned into a full diagnostic knowledge system.

PROJECT 01 · NOSTRAFORMA · SINCE 2009

Commerce systems,
search & platform engineering.

Since 2009, working on the machinery behind an established European design-lighting shop: product data, native search and enrichment, ERP and internal tooling, payments, security, technical SEO, automation, compliance and the operational systems that keep a long-running e-commerce platform moving.

e-commercesearchERPsecuritypaymentsautomation
since 2009 / active
nostraforma.com ↗
PROJECT 02 · nxTLS · 2026 REBOOT

Small HTTPS stack.
Strong opinions.

The successor to nsb2 and TinX: a deliberately small, modern HTTPS stack built from pinned nginx and BoringSSL, with TLS 1.3, post-quantum cryptography, HTTP/2 and HTTP/3 in scope. A clean reboot rather than a continuation of the historical source: upstream stays upstream, policy and integration stay project-owned, and patches must earn their place through measurement. The verified baseline and focused HPACK, QPACK and TLS-record experiments inform decisions; they are not automatically production features.

The philosophy survives. The fork debt does not.

nginxBoringSSLTLS 1.3PQCHTTP/2 + HTTP/3Clang
baseline + experiments
open nxTLS ↗
PROJECT 03 · NGINX EDGE SECURITY

Hardening the edge.
Without confusing signals with trust.

A field guide shaped by years of operating internet-facing systems: protocol lanes, narrow Lua policy, explicit trust boundaries, finite cost budgets and incident-focused observability. Signed cookies prove that steps were completed, not that a client is human. Noisy denials are telemetry; security-relevant successes need their own visibility. New gates must earn their place through measurement.

Not every suspicious request is a security event. Not every valid request is trustworthy.

nginxedge securityprotocol lanesLuatrust policyobservabilityincident detection
living field guide
open edge-security guide ↗
PROJECT 04 · A2 SYSTEM EXPLORER

Diagnosis as a
navigable system.

A living diagnostic and knowledge explorer for the Audi A2: system topology, evidence, failure paths, components, 3D context and a growing model of how the car actually behaves when the wiring decides to become creative.

diagnosticsknowledge graph3DautomotiveAudi A2
active lab
open explorer ↗
02 / BIASES

Opinionated by design.

Not a résumé keyword cloud. Just recurring choices that tend to show up in the things I build.

01

Small surface

Remove what the system does not need. Compatibility is useful until it becomes the product.

02

Current primitives

Modern protocols, explicit crypto policy, current compilers and fewer historical defaults.

03

Inspectable

Prefer systems that can be understood, traced, measured and taken apart without ceremony.

04

Useful first

Tools should solve the actual problem before they become a platform, framework or religion.

03 / ABOUT

Different systems.
Same instinct.

The projects have their own pages now. This is the common thread between them.

Most of this starts with something that already exists.

Commerce platforms, search, nginx, TLS stacks and a twenty-six-year-old car have surprisingly similar failure modes: assumptions accumulate, boundaries blur, and the useful part gets buried under compatibility, folklore or accidental complexity.

I like pulling those systems apart until the real constraints become visible — then making the smallest change that makes them easier to understand, operate or trust.

Build what the system actually needs.
Keep the boundaries explicit. Let everything else justify its existence.

04 / CONTACT

Say hello.

Comments, corrections, strange bugs and interesting rabbit holes are all valid reasons.

No form. No newsletter. No CRM funnel. Just email.

05 / IMPRINT

Imprint.

Legal notice / Anbieterkennzeichnung for xcx.cc.