Research | Saad Ullah Bilal — AI Systems Architect
Research
Staying close to the frontier.
Active research separates consultants who deliver from those who package others' ideas. Staying close to the frontier means knowing which techniques are production-ready and which are still demos.
In Progress
Spatiotemporal Forecasting with Contextual GNNs
A graph neural network framework for high-resolution urban air-quality prediction, combining graph structure learning with temporal attention. The visualization above shows information propagation across monitoring stations.
In Progress
Interpretable Counterfactual Adversarial Examples
A diffusion-based generative framework for studying the robustness–interpretability trade-off in AI systems through semantic disentanglement. The animation illustrates the denoising process.