Arif Dogan — production AI & backend engineer.

I help SaaS teams turn fragile AI prototypes and backend bottlenecks into production systems that can survive real users.

★ Open source

5 tools.
1.2k stars on GitHub.

1.2kgithub · 2024 — present

Production AI integrationTurn a working demo into a reliable feature with evals, tracing, fallbacks, and clean handover.2–4 wks · sprint

Backend rescueDiagnose and fix slow, unstable, or hard-to-change Go/Python systems before they hurt customers.1–3 wks · focused

MVP buildTake a validated product idea from scope to a deployed, production-ready first version.4–8 wks · fixed scope

GoVisualZero-config HTTP request visualizer for Go — debug complex services without leaving the terminal683 ★

FastAPI RadarReal-time monitoring and visualization for FastAPI services in production403 ★

LLMDogContext preparation toolkit that keeps LLM responses reliable for complex inputs82 ★

K9SightKeyboard-driven TUI for debugging Kubernetes workloads — pods, logs, exec, deployments65 ★

GIQAI-enhanced Git workflow assistant that automates code review preparation15 ★

Shipping 100,000 construction PDFs a monthWhat actually breaks at scale in a document processing pipelinearticle

Structuring a managed search engine in RustControl plane, data plane, durable storage: one search engine split across three systemsarticle

Shipping LLM features that survive productionWhy most LLM features break in the real world — and the evaluation layer that keeps them honestarticle

Turning a regulation into testable codeThe method I use to take a statute from primary source to audit-ready detectorarticle

Building a GDPR-compliant backendAn engineer's checklist — the architecture decisions that actually carry legal weightarticle

The EU AI Act for software engineersWhat you actually have to build — obligations translated into system requirementsarticle

The Context TaxWhy AI-assisted coding fails without flowessay

I write about shipping production systems and AI you can trust — the same lens I bring to client work.

MCP in the Wild: Cross-Domain Knowledge Discovery through Multi-Server OrchestrationAn empirical study of multi-server MCP orchestration — 17 real tool calls across 6 servers, a 7-model benchmark, and 5 composition patterns. Preprint.2026

Discovery of TIC 63038821: A New UV Ceti / BY Draconis Variable M-Dwarf Binary from TESS Sector 1Discovered a previously unreported variable star using autoencoder anomaly detection on 15,665 TESS light curves. Flare events, rotational modulation, chromospheric activity confirmed.2026