Applied experiments in AI-assisted software engineering, human-in-the-loop autonomous systems, and AI trust & observability.
Real insights, not academia theater. We focus on what works in production.
Three interconnected research domains that inform all our work.
Where AI actually helps engineers — and where it fails
AI used in coding, testing, infrastructure, DevOps, and QA. We study tooling, workflows, and evaluation — not "vibe coding" or shallow productivity demos.
Fully autonomous systems are brittle. Human-aware autonomy scales better.
Most people skip the human part. We study AI agents with human checkpoints, override mechanisms, escalation design, and cognitive load on humans.
The glue holding everything together
Measuring correctness beyond accuracy. We study drift detection, confidence signaling, auditability, and how to know when AI systems are actually working.
Mapping current AI-assisted workflows in coding, testing, and DevOps.
AI-Assisted SECritical interaction points where human oversight matters.
Human-in-the-LoopInitial metrics for evaluating AI agent reliability.
Trust & ObservabilityNothing is isolated. Nothing is performative.