Optimization, autonomy, and deployable AI systems.
A brief narrative about the research line, the engineering line, and the way they meet in the systems I choose to build.
I am He Hao, a PhD candidate in Electrical and Computer Engineering at Instituto Superior Tecnico and an Autonomous Systems Engineer at SeaPower. My work sits at the boundary between rigorous optimization research and real engineering systems.
On the research side, I work on distributed and nonlinear optimization, multi-agent coordination, continuous-time dynamics, and control-oriented modeling, especially in autonomous surveillance settings where mathematically sound methods still have to survive uncertainty, communication limits, and execution constraints.
On the engineering side, I build deployable software for autonomy and perception, including computer vision runtimes, service-oriented backends, operator-facing interfaces, and bounded AI systems designed to remain usable, inspectable, and reliable in practice.
Optimization & Control
I design distributed and nonlinear optimization methods for multi-agent systems, with emphasis on convergence, decomposition, continuous-time reasoning, and control-oriented formulations that remain meaningful in real operational settings.
Autonomous Systems & Perception
I work on surveillance, planning, estimation, and perception-linked autonomy problems where decision methods must interact with vehicles, uncertainty, sensing, and execution constraints rather than live in isolation.
Platform Engineering
I build operator-facing AI systems with explicit service boundaries, shared runtime components, and workflows that make model outputs inspectable, explainable, and trustworthy in practice.
Theory That Has To Survive Implementation
I care about methods that can still make sense after they encounter interfaces, runtime limits, deployment tradeoffs, and human operators.
Boundaries Before Complexity
I prefer explicit system boundaries, modular services, and clear ownership of responsibilities before adding sophistication to the stack.
Operator Workflow Matters
A technically correct system is not enough if people cannot inspect it, understand it, or act on it with confidence during real work.
Communication Is Part of the System
I document architecture, explain tradeoffs, and connect projects, writings, and experiments so the reasoning behind a system stays visible.
I am most interested in work where theory and deployment have to meet: autonomy, perception, platform architecture, and decision systems that need to remain reliable under real operational conditions.
I am open to selected collaborations across distributed optimization, autonomous systems, computer vision platforms, and operator-facing AI tooling, especially where mathematically grounded methods need to become reliable systems.