I am the lead researcher in TaKaDu , a central event management system for water distribution networks. Our machine learning engine includes a complex transformers engine, and a set of classifiers and physical models that give water utilities an up-to-date online status of their network in terms of leakages, data problems, pump failures, valve breakages etc.
I'm currently working on:
- Asset management model for pipes
- Prediction model for pump failures
- Bayesian model for water supply timeseries
- Water supply data cleaner
- Hybrid statistical and physical model of a water distribution network
My interests include : Timeseries analysis, anomaly detection, recommender systems and complex networks
I was also one of the initiators of the field of climate networks
I previously worked on : El-Nino and La-Nina statistical predictions, teleconnections, double diffusion, ocean dynamics and instabilities and stochastic transport