AI-Augmented Engineering Sprint
Situation
After leaving ezCater, I used a sabbatical not as a break but as a return to hands-on engineering. The goal was to build at high velocity across a wide range of product and platform problems using AI-assisted development.
This was deliberate: I wanted to test whether a senior engineering leader could maintain technical currency and building speed by deeply integrating AI into the development workflow — and whether that experience would produce transferable judgment about AI-augmented engineering practices.
Decision
I committed to building across product, workflow, automation, and tooling domains. Every project started from a real customer problem, not technology for its own sake.
The result: 9,628 GitHub contributions in the last year across 100+ repositories. This is not volume for its own sake — it reflects repeated reps across varied problem spaces, each producing judgment about what works, what breaks, and what AI-assisted development is actually good for.
Risk
The risk of a sabbatical build sprint is diffusion. Building 100+ things can mean finishing nothing and learning nothing transferable. I managed this by treating each project as a judgment exercise: what’s the real problem, what’s the minimum viable approach, what would break in production, and what did I learn.
There’s also a perception risk: some organizations view sabbatical work as hobbyist activity rather than professional development. I accepted this because the work speaks for itself in both volume and variety.
Change
The sprint produced:
- A curated set of projects spanning SaaS products, developer tools, workflow automation, and AI agent frameworks — see the Build Lab for representative examples.
- Transferable judgment about AI-assisted development: where it accelerates, where it misleads, where human judgment remains critical.
- Demonstrated ability to operate as a hands-on builder while maintaining the systems thinking and product judgment of a senior engineering leader.
What This Demonstrates
- Technical currency: Not theoretical — 9,628 contributions across 100+ repositories in one year.
- AI-forward practice: Deep integration of AI into the development workflow, not as a novelty but as a core capability.
- Product judgment at speed: Each project starts from a real problem, not a technology demo.
- Breadth with depth: Varied domains (product, platform, tooling, automation) with enough depth in each to produce real judgment.