Ronan Doorley, Developer in Dublin, Ireland
Ronan is available for hire
Hire Ronan

Ronan Doorley

Verified Expert  in Engineering

Regression Developer

Location
Dublin, Ireland
Toptal Member Since
December 4, 2018

Ronan has several years of experience in data science and machine learning with Python, R, and MATLAB. He has a passion for developing validated models of human behavior and complex systems. Currently, he works with the City Science group at the MIT Media Lab where he develops predictive models of individual behavior such as activity scheduling, location choices, and modes of transportation.

Machine LearningMATLABPython 3NumPySciPyPandasProbabilistic Graphical ModelsRFlaskNetworkXPySparkBokehMySQLJupyterGitRStudio ShinyRandom Forest

Availability

Part-time

Preferred Environment

GitHub, Jupyter, Spyder, Anaconda, MacOS

The most amazing...

...model I’ve developed simulates a human population from an urban plan and predicts their daily activity schedules and transport mode for each activity.

Work Experience

Mobility Modeling Engineer

2016 - PRESENT
MIT Media Lab
  • Developed AI for agents in urban simulations using a cascade of Bayesian Network and Random Forest models calibrated with survey data.
  • Built a Bayesian network model in Python for generating traffic predictions in the country of Andorra; it was based on a combination of cell phone records and traffic counter data.
  • Created also the front-end visualization for the traffic project using Web Sockets and Mapbox GL.
  • Developed a Poisson process model in R to explain the formation of dense clusters of social activity in cities based on the physical urban features.
  • Deployed a Python Flask API which uses a pre-calibrated discrete choice logit model to predict mobility patterns in response to an input describing changes to land use.
  • Used the spatiotemporal data from app usage to develop a discrete choice model in Python to describe how people choose which amenities to visit.
  • Implemented neural networks in Python Keras to model proxy metrics for urban vibrancy based on Google Street View images.
Technologies: Bayes Net Toolbox (BNT), Random Forests, Analysis, Classification, Scikit-learn, Flask, Python

Doctoral Researcher | Teaching Assistant

2013 - 2016
Trinity College Dublin
  • Developed a mathematical framework for quantifying the benefits and risks of walking and cycling for transport in urban environments.
  • Designed and built a mobile pollution sensing node on the Arduino platform and used it to characterize the exposures of pedestrians and cyclists to various pollutants in Dublin.
  • Constructed a game theoretical model in MATLAB describing how people change their transportation behaviors in response to changes in the cycling infrastructure.
  • Created a genetic algorithm in MATLAB to find the optimal design of a cycle network, considering the expected behavioral responses and the resulting health and environmental impacts.
Technologies: Arduino, R, Python, MATLAB

Analyst

2010 - 2011
Accenture
  • Created functional designs and tested web applications for clients in the financial sector.
  • Worked closely with client resources to determine and document their requirements in areas such as data capture, system functionality, pricing, risk acceptance criteria, automatic document generation, and more.
  • Developed product prototypes using Axure and led teams of about ten developers in building fully functional web applications based on these prototypes.
  • Thoroughly tested the product during development, mainly using automatic testing scripts.
Technologies: Selenium, Axure

CityScope MoCho

https://www.media.mit.edu/projects/mobcho/overview/
This project produced a tool which allows urban planners and community members to interact with a physical model of a city district and see how different urban designs would influence the mobility patterns.

I built a Flask app which reads the state of the district design from an API and uses a discrete choice logit model to predict the changes in travel patterns and resulting environmental impacts at a regional level. The analysis results are exposed through the Flask API.

Reversed Urbanism

This project developed a regression model to explain and predict where dense clusters of urban activity form, based on the physical characteristics of the area. I first analyzed geolocated telecom data, which was obtained in the country of Andorra, to identify clusters of activity characterized in terms of their size, persistence, and diversity. I then built a lasso regularized multivariate linear regression model in R to order to identify associations between the formation of these clusters and various discrete urban features.

Dynamic Traffic Prediction in Andorra: a Bayesian Network Approach

https://www.media.mit.edu/projects/traffic-andorra/overview/
In this project, I built a Bayesian network in Python to predict trips made and traffic congestion in Andorra based on a combination of geolocated telecoms data and a small sample of traffic counts.

Languages

Python 3, R, Python, JavaScript, HTML, C++

Libraries/APIs

Scikit-learn, NumPy, SciPy, Pandas, Matplotlib, Mapbox GL, NetworkX, PySpark, D3.js, Leaflet, jQuery, Keras

Tools

MATLAB, Spyder, GitHub, Bayes Net Toolbox (BNT), Jupyter, Git

Other

Machine Learning, Research, Statistics, Regression, Classification, Software Development, Probability Theory, Probabilistic Graphical Models, Neural Networks, Axure, Analysis, Bokeh, WebSockets, HTTP, Ajax, Experimental Design, Random Forests

Frameworks

Flask, Selenium, RStudio Shiny

Paradigms

REST, Data Science

Platforms

MacOS, Anaconda, Arduino

Storage

MySQL

2016 - 2017

Fulbright Scholar in City Science

Massachusetts Institute of Technology | MIT - Cambridge, MA, USA

2013 - 2016

PhD Degree in Civil, Structural and Environmental Engineering

Trinity College Dublin - Dublin, Ireland

2006 - 2010

Bachelor's Degree in Mechanical Engineering

Trinity College Dublin - Dublin, Ireland

SEPTEMBER 2015 - PRESENT

Data Manipulation at Scale: Systems and Algorithms (MOOC)

University of Washington via Coursera

JULY 2014 - PRESENT

The Data Scientist's Toolbox (MOOC)

Johns Hopkins University via Coursera

JULY 2014 - PRESENT

Statistical Inference (MOOC)

Johns Hopkins University via Coursera

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring