Software architect · AI-agent engineer
I design systems meant to last past their second year, and I build my own products on them.
I own the architecture on each of these and write most of the code.
| Product | What it does | |
|---|---|---|
| SEOG | AI local-SEO platform — Google Business Profile management, map-pack rank tracking on a geo-grid, review sync with published replies, competitor and AI-visibility intelligence, and reports. Also usable straight from an AI agent's terminal. | seog.ai |
| VidVibe | AI video platform — generate, edit and publish short-form video to multiple channels from one place, with brand kits, consistent characters and automated flows. | vidvibe.io |
| Terminal Skills | Marketplace and open library of AI-agent skills — portable skill files that drop into the major coding agents, with a CLI to install them and automated validation of every submission. 1,000+ skills, 173K+ installs. | terminalskills.io |
| Kompot | Multi-tenant CRM with an AI layer on top — document and receipt parsing, voice interface, and routing of AI tasks across providers with automatic failover. | kompot.ai |
| Skyline | Self-hostable deployment platform — "your own Vercel" on infrastructure you control. One panel takes a site from repo to live: push-to-deploy, build logs, custom domains and DNS, environment variables, asset CDN and instant rollbacks. | skylinesite.com |
| OmniVerbis | Realtime speech and document translation. Conference mode — the audience scans a QR code and each listener reads or hears the talk in their own language — plus a two-person conversation mode with one phone on the table. Captions land about a second after the speaker pauses. | omniverbis.com |
Also built: Card Forge, a multi-tenant SaaS for digital smart cards with AI image generation in the editor, and 3D Drone Tour, which turns drone footage into a navigable 3D property walkthrough in the browser.
Architecture. Multi-tenant SaaS, event-driven backends, monorepos with shared contracts, Postgres/Mongo data modelling, and the boring discipline that keeps them maintainable — typed boundaries, migrations, real CI.
AI-agent engineering. This is where most of my time goes now: agent orchestration and fleets, MCP servers and tool design, RAG with vector search, self-hosted inference next to hosted models with failover routing, and OpenTelemetry-based observability so agent behaviour is measurable instead of vibes.
Infrastructure. Docker, Cloudflare Workers, Coolify, GPU boxes for local inference, LGTM stack for metrics/logs/traces. I deploy what I build and I'm on call for it.
Product. Payments, auth, i18n and RTL, SEO, transactional email, and the unglamorous integration work that decides whether a product actually ships.
- TerminalSkills/skills — open library of AI-agent skills.
SKILL.mdfiles that drop into Claude Code, Codex, Gemini CLI or Cursor and teach an agent a task. - seog-ai/local-seo-manual — a free, hands-on manual for local SEO: how Google ranks businesses in the map pack, 49 chapters, real labs and a probe-verified field reference.
Languages
Backend
Frontend
AI / agents
Data
Infra & observability
Open to partnerships, product collaborations and interesting architecture problems.
Book 30 minutes →