He Hao
Autonomous Systems Engineer · PhD candidate
Optimization, control, and software for autonomous and AI-enabled systems.
At SeaPower, I build deployable AI-enabled systems for autonomous platforms, drawing on doctoral research in distributed optimization for multi-agent systems at Instituto Superior Técnico.
Wildfire Surveillance Autonomy
From Fast-Mixing Coverage to Smooth Risk-Aware Aerial Search
ML Competition Portfolio
Selected Medal Cases, Retrieval Work, and Public Artifacts
ASV Navigation Stack
Planning-Stage Autonomy Architecture for Surface Vessels
Quadrotor Surveillance: Local Peak Seeking and Smooth Coverage Control
Technical Deep DiveDynamics of Optimization: From Faster Iterations to Faster Settling
Technical Deep DiveLumbar Spine Classification: From One Global Model to Split Specialists
Engineering StrategyDesigning the Markov-Matrix: Why Adapting a Random Walk Fell Short
Technical Deep DiveDistributed Optimization
Gradient tracking, ADMM, and asynchronous methods for multi-agent objectives over communication networks.
Nonconvex Analysis
Convergence guarantees for nonconvex problems — saddle point escape, stationary point characterization.
Autonomous Systems
Formation control, consensus-based coordination, and path planning for marine and aerial autonomous vehicles.
ML & Optimization Interfaces
Connecting optimization theory to ML training dynamics, hyperparameter search, and differentiable programming.
Distributed Optimization
Gradient tracking, ADMM, and asynchronous methods for multi-agent objectives over communication networks with convergence guarantees.
Nonlinear Analysis
Convergence analysis for nonconvex optimization — Lyapunov methods, KKT conditions, stationarity, and rate bounds.
Autonomous Systems
Sensor fusion, path following, and distributed coordination for autonomous surface and aerial vehicles.
ML & Optimization Interfaces
Connecting optimization theory to ML training, hyperparameter search, differentiable control, and learning-to-coordinate.
Lab Highlights
Agentic Assistant for a CV Operator Platform
An active systems investigation into how an LLM assistant should sit inside a FastAPI + React + YOLO platform: not as a chatbot bolted onto the UI, but as a tool-using orchestration layer over inference, catalog, and stream services with approvals, memory, and runtime guardrails.
Smooth Coverage over Unknown Utility Fields
A local-surveillance lab thread on turning only local gradient information into smooth quadrotor motion without relying on a globally known utility map.
Split Specialists for Lumbar MRI Classification
A competition lab note on decomposing a heterogeneous medical-imaging task into smaller specialist models instead of forcing one global classifier to absorb everything.
Key Milestones
Joined SeaPower as an Autonomous Systems Engineer, building autonomous platforms.
Published key doctoral research outputs on gradient descent dynamics, stochastic coverage, and hybrid optimization for drone surveillance.
Worked as a part-time Data Scientist at Guangyuan Keyou and reached Kaggle Expert tier.
Began full-time PhD research under competitive FCT funding at Instituto Superior Técnico for wildfire surveillance in drone swarms.
Started the Ph.D. in Electrical and Computer Engineering at Instituto Superior Técnico.
Awarded Scholarship Within the Framework of R&D Projects and Institutions (BL204/2018).
Open to research collaborations, consulting, and conversations at the intersection of optimization, autonomous systems, and machine learning.