Ibrahim Faisal

I'm an AI solutions architect based in Sydney. Most of my work sits at the intersection of autonomous AI systems, cloud-native data architecture, and the gap between what gets designed and what actually ships. That last part — architecture to delivery — is where I spend most of my attention.

In practice, this means I spend a lot of time on the systems around the model: the control plane, the evaluation layer, the orchestration logic that determines whether an agentic system behaves predictably or quietly drifts. The model is rarely the problem. The infrastructure around it usually is. I've written about this more directly in the lab notes, but the short version is that agentic systems are engineering problems, not prompt engineering problems.

Before AI systems became the dominant theme, I worked across data engineering and platform architecture — designing pipelines, building cloud-native data platforms, and helping teams close the distance between architecture on paper and something running in production. That background shapes how I approach AI work now. I care about observability, failure modes, and whether the thing can be debugged when it breaks. Those concerns don't disappear because the compute is probabilistic.

Deadlock Labs is where I keep the ideas that don't fit inside a work engagement. It's a public lab for product explorations, AI experiments, and problems I want to think through seriously. Some of the ideas are product-shaped. Some are research-adjacent. I publish them openly because ideas that stay private tend to stay unfinished.

Outside of professional work, I follow Arsenal in the Premier League, and track Pep Guardiola and Haaland wherever they play. Cricket means Tests — one day is slowly becoming irrelevant, and T20 is a different sport with the same name. I watch a lot of film, mostly festival releases and contemporary drama, and have been shooting photography long enough to have preferences about it without being particularly systematic.

This site is where those threads come together.