Machine Learning Developer in Barcelona, Spain

One of my last projects at Jaguar Land Rover.

Go I-PACE helps customers understand the potential cost savings of going electric compared to their existing vehicle. would-be buyers. The app estimates how I-PACE would fit into your life based on personal journey data.

The Go I-PACE app captures journey data to calculate potential cost savings, show how much battery would be used per trip and tell users how many charges they would need in a week if they were driving the I-PACE.

Calculates the range expected from a full charge based on your vehicle use, the number of charges required in a typical week and how frequently you would need to top up mid-journey.

It can also distinguish between different modes of transport to make sure it collects accurate data, even prompting users to confirm that individual trips were made by car for unusual routes – for instance on journeys made by cycling rather than behind the wheel.

Responsible for the delivery of the data analytics components of a car and mobile application to offer fully personalized driving experience to Jaguar Land Rover customers and help prevent accidents by reducing driver distraction.

Main responsibilities and goals involved:

● Define and implement scalable real-time workflow to load data, quality management, and distribution across various system using Big Data technologies on Amazon Web Services.
● Contribute to the software development lifecycle including the analysis, architecture, design, implementation, and QA.
● Hands-on work directly implementing complex machine learning solutions using Natural Language Processing, recommender systems, neural networks and/or deep learning.
● Write technical documentation and presentation of results to technical and non-technical stakeholders.

Scala package to process time series from different sensors with Spark.

Processing time series collected from different sensors poses several challenges as a result of data may not be aligned or have the same time sampling. Writing data queries can be quite hard for data scientists because data cannot be expressed in a tabular form.

This library makes it easy to write queries with this kind of datasets.

Analysis of daily driving patterns of Jaguar Land Rover customers:

- Fuel consumption
- Daily in-car time
- Commute schedule
- Regular routes
- Total distance
- Journey duration
- Driving style
- Refuelling events
- Phone call patterns
- Heated and cooled seat usage
- Phone call pattern
- Radio stations

This challenge will stimulate research on network-structure learning from neurophysiological data, including causal discovery methods.

The goal of the data science competition was to predict the directed connection between 1000 neurons based on their time series of the activity.

My solution involved a mixture of several features such as correlation, mutual information, partial correlation, spectrogram, and frequency analysis.

This competition challenges you to identify when a hand is grasping, lifting, and replacing an object using EEG data that was taken from healthy subjects as they performed these activities. A better understanding of the relationship between EEG signals and hand movements is critical to developing a BCI device that would give patients with neurological disabilities the ability to move through the world with greater autonomy.

My solution involved a deep neural net developed with Python using Theano and Lasagne.