Dénes Tóth, Ph.D.

I worked as the lead developer of a full-stack SaaS analytic dashboard application for an Australian organization (RADII ProLearning, Radii.org/prolearning). The project includes extensive data integration and preprocessing steps, automated real-time statistical and predictive modeling, interactive visualization, and reporting.

The project grew out of a smaller-scale task, including the integration and cleansing of educational data collected from primary schools in Australia. The task was very challenging due to the total lack of documentation and extreme variability of data sources and data quality. Fun fact: I was first hired to build structural equation models on the data—it became evident very quickly that this would not be possible without the appropriate input.

Now the system is a SaaS web application hosted in Azure with lots of capabilities not covered by any other contemporary solutions on the market: the school user can run complex analyses and create informative reports intuitively without any knowledge about statistics. The software is in beta and shows great potential, also supported by the Microsoft for Startups program.

A consultancy working with a large multi-depot wholesale company delivering goods to ~6,000 clients per week approached me for advice on optimizing the logistics network of their client. Specifically, the client wanted to know how many drivers they need for a highly reliable and still cost-efficient delivery network.

At that time, I knew almost nothing about vehicle routing problems. However, in one month, I reviewed the literature, chose an open-source Java library, and despite being totally new to Java, tweaked the code to meet the need of the actual problem, wrote the R scripts which preprocessed the data, and ran the required simulations.

In the end, the client was so satisfied they asked me to advise them on further optimization possibilities, e.g., improve the depot-assignments or visiting days of their customers.

Later, the client decided to develop a highly customized vehicle routing software that can handle their specific demands. I developed and tuned the algorithm of the software and helped the client identify the functional requirements. The software was implemented in Scala, to which I contributed high-level pseudocode and several Java prototypes. The system is in production since 2017.