Engineer · AI Integration · Forward-Deployed
Full-stack engineer who ships production AI integrations
I'm Ryan Dierickx, a Power BI engineer at Microsoft and the architect of Starbucks' Digital Production Manager system. I build with Nuxt, Vue, and TypeScript, and I put Claude, Codex, and local LLM orchestration into workflows people actually use.
Most candidates have web or AI. I ship both, at enterprise scale and as a solo founder.
Now
Microsoft
Power BI engineer building interactive visuals used by thousands of businesses daily
Previously
Starbucks
Created and led the Digital Production Manager system
Founder
RND Tech & Design
Custom web + AI builds with true client ownership
What I Bring
Enterprise engineering discipline, hands-on AI integration experience, and the product sense to pick the problems worth solving.
AI Integration & Agents
Wiring Claude, GPT, and local models into product workflows: RAG, structured output, evals, cost and latency tuning, and agent orchestration with guardrails.
Full-Stack Web Engineering
Nuxt 4, Vue 3, TypeScript, Tailwind. Static-generated, performance-first builds that ship fast and stay maintainable. Production sites, portals, and internal tools.
Data & Visualization
Power BI engineering at Microsoft scale, on interactive visuals that thousands of businesses use daily. I'm comfortable turning messy operational data into decisions.
Forward-Deployed Delivery
I sit with the customer, learn the domain, and ship working software against their actual data, not a slide deck. Equally at home solo or embedded with a team.
Production AI, Not Demos
Two of these are running on this site right now. Each one pairs a model with hard constraints, validation, and a human in the loop.
Live on this site
AI-Powered Website Audit
Problem
Prospects wanted to know what was wrong with their site before committing to a project, but manual audits don't scale.
Approach
Built an intake flow that feeds page data into an LLM analysis pipeline with a constrained, schema-validated output, then renders an interactive strategic report. Human review stays in the loop before anything ships to the client.
Nuxt SSGLLM pipelineStructured outputNetlify functions
Live on this site
Blueprint Planner (SOW Generator)
Problem
Scoping conversations were slow and inconsistent, and prospects bounced before a quote existed.
Approach
A guided, conversational planner that turns a few answers into a structured statement of work and preliminary scope. The model drafts and I make the calls, so judgment stays with me and the busywork doesn't.
Vue 3TypeScriptPrompt designStateful forms
Personal R&D
Local AI Orchestrator
Problem
Cloud-only AI tooling means giving up privacy, portability, and control over your own workflows.
Approach
A local-first system that orchestrates multiple coding agents (Claude, Codex, Copilot, Aider) over a version-controlled knowledge base, with durable memory, repeatable jobs, and private context that never leaves the machine.
Local LLMsMulti-agent orchestrationCLI toolingMarkdown knowledge base
Architecture walkthroughs and a code review of any of these are available on request.
Stack & Tooling
TypeScriptVue 3Nuxt 4Tailwind CSSNode.jsClaude / Anthropic APIOpenAI / CodexGitHub CopilotLocal LLMs (Ollama)RAG & embeddingsPower BI / DAXNetlifyStatic Site GenerationSanity
Available For
Open to the right full-time role, and taking on a limited number of fractional and project engagements.
Forward-Deployed / Solutions Engineering
Embedded with your customers or product team, shipping AI integrations against production workloads.
Fractional AI Engineering
An ongoing technical partner for teams that need AI capability without a full-time hire.
Integration Sprints
Fixed-scope engagements to wire a specific AI workflow into your stack and prove it out.
Let's Build Something That Ships
Hiring, scoping an integration, or just want to compare notes on shipping AI in production? I'd like to hear about it.
