Khalil Al Sayed
PhD Researcher in Artificial Intelligence.
Specializing in autonomous HVAC control and smart building energy management systems using Deep Reinforcement Learning.
About
Bridging Theory & Application.
I hold a PhD in Artificial Intelligence, specializing in Deep Reinforcement Learning (DRL) for Building Energy Management Systems (BEMS). My research focuses on optimizing energy consumption in complex systems using tools like EnergyPlus and PyTorch.
With a strong foundation in applied mathematics and data science, I have authored 6 publications (including IEEE conferences) on optimal control. I enjoy bridging the gap between theoretical AI models and real-world industrial applications.
Expertise
Technical Skills
Selected Work
Research & Engineering.
EnergyPlus-Gym Framework
A custom Reinforcement Learning environment connecting EnergyPlus simulations with OpenAI Gym. The backbone for training advanced HVAC control agents.
ATMO Analytics Dashboard
An interactive dashboard analyzing correlations between socio-economic indicators, home-work mobility, and CO₂ emissions.
Attention Makes HVAC Control More Efficient
IECON 2025. Investigating the impact of attention mechanisms in Deep Reinforcement Learning agents. By focusing on relevant state features, this approach significantly improves energy efficiency and convergence speed in complex building environments.
On-Policy vs. Off-Policy HVAC Control
IEEE UEMCON 2025. A critical comparative study of PPO (On-Policy) and SAC-Gumbel (Off-Policy) algorithms. The research evaluates sample efficiency, stability, and thermal comfort compliance in high-fidelity EnergyPlus simulations.
DuoBrain-HVAC: Hybrid Autonomous Agent
TELFOR 32nd Forum. Introducing a hybrid control architecture that combines the robustness of rule-based systems with the adaptability of Reinforcement Learning. This "dual-brain" approach ensures safety constraints while optimizing for energy savings.
RL for HVAC Control: A Technical Review
Journal of Building Engineering. A comprehensive survey of state-of-the-art RL methodologies for building energy management. This work synthesizes findings from recent studies to identify best practices and future research directions.
SVD Text Mining Engine
A high-performance C++ implementation of Singular Value Decomposition (Golub-Kahan + QR algorithms) designed for text-mining applications and large-scale matrix operations.
Academic
Publications
Attention Makes HVAC Control More Efficient
IECON 2025 (IEEE Industrial Electronics Society)
Investigating attention mechanisms in DRL agents for improved energy efficiency.
Read PaperComparing PPO and SAC-Gumbel in EnergyPlus
IEEE UEMCON 2025
A comparative study of policy optimization algorithms for building energy simulation environments.
Read PaperReinforcement Learning for HVAC Control: A Technical Review
Journal of Building Engineering
A comprehensive technical review of current RL methodologies applied to HVAC control systems.
Read PaperDuobrain-HVAC: A Hybrid Autonomous Agent
TELFOR 32nd Forum
Presented a hybrid agent architecture combining rule-based and learning-based approaches.
Read PaperRL for Optimal HVAC Control: Theory to Application
ResearchGate Publication
Bridging the gap between reinforcement learning theory and real-world HVAC control applications.
Read PaperContact
Let's collaborate.
I'm always interested in discussing research collaborations, smart energy systems, and AI optimization. Feel free to reach out.