Ongoing projects I’m leading:

Dash Sylvereye: A WebGL-powered Library for Dashboard-driven Visualization of Large Street Networks

Dash Sylvereye is a new library to produce interactive visualizations of primal road networks on top of tiled web maps that can be consumed in modern web browsers. Dash Sylevereye can render large interactive road graphs in commodity systems by exploiting WebGL; and can be used in combination with Plotly Dash to generate rich dashboards around geo-spatial road data. Dash Sylvereye is the product of a collaboration between the Center for Research in Geography and Geomatics (Mexico) and the University of Twente (the Netherlands).

Frigate: An End-to-end Stream Computing Platform for Self-adaptive Vehicle Route Planning

Frigate is an end-to-end platform for processing streams of Floating Car Data (FCD) (e.g. raw GPS positions and vehicle speeds) to learn vehicle route plans in real-time by means of a stream-based Reinforcement Learning approach. This work strives to produce an end-to-end solution for the vehicle route planning problem, based on a novel Q-routing algorithm that learns in real-time and in a distributed manner to predict travel times between junctions in a road network. Frigate is the product of a collaboration between the Center for Research in Geography and Geomatics (Mexico) and the University of Twente (the Netherlands).

Whistler: A Cloud-native SNA Platform for Distributed Processing of Large Tweet Datasets

Whistler is a cloud-native application, PaaS platform for distributed social network analysis (SNA) on large tweet datasets. Through its web interface, Whistler exposes a variety of SNA techniques for tweet analysis, including sentiment and link analysis. Under the hood, Whistler leverages the benefits of container-based virtualization to provide distributed and elastic data processing capabilities to both Whistler’s interactive web dashboards and WhistlerLib, a new distributed computing library for programatically processing tweet datasets on top of Whistler.

Weaverlet: A Component-driven Web Framework for Compositing Complex Dashboard Visualizations

Weaverlet is a server-side component-driven web framework to build multi-page web applications with coordinated and interactive dashboard visualizations written entirely in Python. No JS/HTML/CSS/templating required. Weaverlet is designed around the concept of Weaverlet Components: classes that encapsulate the complexities of the layout and callbacks of one or more Dash components that together perform a single higher-level UI function, presenting them as a self-contained, composable, and reusable UI component to other Weaverlet Components. Weaverlet Components can be nested and communicate to each other by means of signals. In this way, a complex multi-page dashboard visualization application can be entirely built out of nested Weaverlet Components that together form a Weaverlet Component Directed Acyclic Graph.

Albatross: A Cloud-native PaaS Platform to Deploy Dashboard Visualization Applications for Big Data Analysis

Albatross is a cloud-native PaaS platform for testing, deploying, and scaling web applications built around dashboard visualizations written in Python for Big Data analysis. On the visualization end, Albatross exploits the Weaverlet framework to enable building of large Dash web frontends entirely in Python, and provides a distributed hub for Weaverlet dashboards, scaling to many dashboard apps. On the Big Data end, Albatross abstracts the user from the complexity of configuring and running a stack of services from the Python Big Data ecosystem, such as PySpark, Dask, AirFlow, Faust, and Celery. In addition, Albatross deploys dashboard applications on top of Docker Swarm, facilitating elastic deployment and scaling on IaaS cloud clusters.

Past projects I’ve led:

Safari: Situational Awareness Framework for Risk Ranking

Safari is a situational awareness framework that exploits different perspectives of the same financial data and assigns risk scores to entities (e.g. payment documents) to improve false positive ratios and assist the identification of fraudulent activity in large unlabeled financial data. Safari is the product of a collaboration between the MIT Geospatial Data Center (US) and Accenture (Ireland).