Build new projects
From initial idea and architecture to a working, deployable system with infrastructure, observability and documentation.
I combine long-term DevOps and infrastructure experience with ChatGPT, Codex CLI, Git, RFC-driven delivery, observability, and session evidence so AI-assisted engineering stays practical, auditable, and under human control.
My background is in Linux systems, infrastructure automation, networking, virtualization, containers, CI/CD and platform operations. I now apply AI-assisted engineering as a reviewed repository workflow: define architecture and constraints, break work into RFC-sized tasks, use Codex CLI for bounded implementation, validate the result, preserve sanitized session evidence, and take responsibility for integration, security and operations.
From initial idea and architecture to a working, deployable system with infrastructure, observability and documentation.
Understand unfamiliar code and infrastructure, identify operational risks, stabilize the environment and establish a maintainable workflow.
Use ChatGPT and Codex as engineering tools for architecture, RFC preparation, repository work, validation and iterative delivery.
Linux, Docker, Terraform, Ansible, GitLab CI/CD, monitoring, networking, virtualization and reproducible environments.
A controlled execution and monitoring platform for predefined Docker-based jobs with live logs, PostgreSQL-backed history, lifecycle management and observability.
A Dockerized public-data collection component for Google Maps search results, structured output and execution metrics, prepared for controlled job execution workflows.
An infrastructure automation project showing a Terraform to Hyper-V to generated Ansible inventory to Dockerized Ansible workflow for reproducible lab environments.
An engineering journal and operating record for AI-assisted delivery: RFCs, Codex CLI handoffs, Fast Bootstrap, session exports, Git history, review notes and human validation.
A central Prometheus, Loki and Grafana stack that grew from GitLab monitoring into shared observability for applications and infrastructure.
A legacy hardware lab for Debian, Home Assistant, kiosk experiments and battery/EC investigation with AI-assisted troubleshooting.
Prompts, generated files and iterative technical work.
Version control became essential after repeated AI-generated changes occasionally broke working code.
I experimented with OpenAI API, prompt history and automatic file updates committed to Git.
The workflow shifted from receiving code to delegating bounded implementation tasks in real repositories.
ChatGPT prepares RFCs, Codex CLI works in Git, tests and validation run, sanitized session reports return for independent review, and the next RFC corrects course.
Fast Bootstrap, declared repositories, session metadata and export tooling make local Codex runs auditable instead of one-off chat experiments.
Built a local RFC runner around explicit repositories, Fast Bootstrap, safe SSH/Git assumptions, sanitized session exports and reviewable return packages.
Moved monitoring into the site repository: Nginx JSON logs, Grafana Alloy, Loki delivery, log rotation, provisionable dashboards and GeoIP enrichment.
Used an Nginx/Prometheus regression to strengthen the workflow: verify the requested function and the neighbouring paths that depended on old assumptions.
Combined Debian administration, conservative battery testing, ACPI/EC analysis and firmware evidence to repurpose old hardware safely.