Publications

Deep reinforcement learning policies for autonomous ultrafiltration control that maximize net water production under operating and safety constraints, evaluated against conventional rule-based backwash scheduling.

An end-to-end machine learning pipeline for ultrafiltration plants: late-fusion forecasters for net water production, model benchmarking across horizons, and anomaly monitoring over operational KPIs.

Transformer-based multi-style networks for predicting human motion in industrial human–robot collaboration, adapting style-conditioned architectures to short-horizon pose forecasting.

A multistage stochastic programming formulation for pharmaceutical supply chains under disruption, balancing inventory cost against service level across demand and capacity scenarios.