I take ideas from their first rough sketch to dependable, production-ready software. My work sits at the intersection of applied AI, automation, integrations, and full-stack product delivery—usually where a messy real-world workflow needs to become a clear, useful system.
Based in Ottawa, Canada. I care about privacy-conscious architecture, explicit human approval for consequential actions, and software that earns trust after the demo.
| Focus | What that looks like |
|---|---|
| Idea → production | Shape ambiguous concepts, prototype quickly, validate the hard parts, and ship the complete workflow. |
| Applied AI | Grounded assistants, agentic workflows, structured extraction, and reliable LLM guardrails. |
| Automation & integration | Turn repetitive work and disconnected tools into observable, maintainable systems. |
| Creative engineering | Build at the meeting point of code, interaction, sound, and human expression. |
- Arpa Meal Planner — a full-stack planning system that connects recipes, weekly schedules, pantry inventory, grocery deduction, and AI assistance.
- GEDS Explorer — a privacy-conscious, explainable research workspace built from official Government of Canada directory sources.
- Top Notch — a local-first macOS workflow surface for music, lyrics, clipboard history, and lightweight notes.
- ApplAI — a suggestions-first career workflow for reviewing AI-assisted CV changes before approved exports.
- Air Controller — a low-latency webcam-to-MIDI controller for shaping sound through hand gestures.
I also contribute to Codex Router, an independent community project for safely connecting Codex to external model providers.
AI & orchestration · LLM workflows, RAG, structured outputs, local models, human-in-the-loop systems
Product & web · TypeScript, React, Node.js, Express, FastAPI, Vite
Data & automation · Python, SQLite, APIs, integrations, browser automation
Native & creative · Swift, SwiftUI, AppKit, Kotlin, MIDI, MediaPipe
idea → frame the problem → prototype → test the risky parts → ship → learn
- Keep the architecture proportionate to the problem.
- Make sensitive or irreversible actions explicit and reviewable.
- Prefer evidence from working software over impressive-sounding claims.
I produce music, write songs, and play guitar, bass, keyboards, and vocals. Studio work taught me the same lesson software keeps teaching me: the smallest details often decide whether the whole thing feels right.
İyi yazılım da iyi müzik gibi—ritim, yapı ve dikkat ister.
If you're building an AI product, an automation-heavy workflow, or something creative that needs solid engineering, take a look at my work or connect with me on LinkedIn.