Project volunteer Orfeas Kypris
Orfeas has a *cough* electrifying background in electromagnetism, machine learning, and software engineering. He’s already done some ace volunteering as part of our community and beyond, working on tools that support the environment through geolocation.
Tell us a bit about yourself and your data background
My name is Orfeas, and I am an electrical engineer with a PhD in applied electromagnetism. I am a Senior Software Engineer at a top-five music industry company. I specialise in signal processing and machine learning applications, and currently build, integrate, and scale internal and customer-facing software products. I am passionate about prototyping, integrating, and scaling software solutions and data ecosystems.
My trajectory into software and data was not linear — I started out as an electrical engineer. During my Bachelor’s degree, I became interested in the physics of magnetism.
My interest in data and how it can be manipulated to extract information came from developing magnetic sensors during my Masters and PhD. I designed and built sensors to measure material properties — from hardware design all the way to developing the signal processing code that extracted features from the measured data.
After a post-doc in a similar field, I took part in a data science bootcamp where we worked on a real-life data science project with a partner company. I became more interested in machine learning when I saw its potential for solving otherwise intractable problems. I joined a startup as a software engineer, and at the same time was reading all the machine- and deep-learning materials that I could get my hands on.
I then joined a robotics company where I led computer vision projects, while at the same time studying various books on statistics, machine learning (ML) and taking courses on Coursera and DataCamp (which I still do to this day).
What is your daily data work like and what tools do you use?
At the moment, I focus on making sure our software stack runs robustly and at scale. I mostly use Python and the pydata ecosystem, use C++ to optimize performance-critical code, Docker, Kubernetes, AWS, develop REST APIs, and am currently looking into using Airflow for running ML pipelines at scale.
What do you wish you’d known when you started to get into data?
The unknown unknowns for me back when I started getting into software and data were version control and unit testing. Unfortunately, I had to learn the hard way that untested code may very well work in your local environment, but once it gets deployed, bugs will eat it alive. Unit- and end-to-end testing really helped me gain confidence in my output, and I’ve never looked back.
What is a data project that inspires you?
A project I am passionate about is not directly related to data analysis, but data collection: I recently built a mobile application for tracking reforestation efforts led by a global non-profit, the International Tree Foundation. This application was built around the nonprofit’s need for accurately monitoring the planting of each individual tree, which includes recording the species, along with photos and geolocation.
It was a pleasure for me to build this tool, as I know how the restoration of degraded forests can be a substantial ally in the fight against climate change. As data collection ramps up, there will be ample opportunity to leverage the data for analytics purposes, helping extract valuable insights that could assist reforestation efforts.
Did you always think you were going to go into data?
I grew up thinking I was going to be a professional musician, and then an engineer working in green technologies. Between working full-time jobs and volunteering, I somehow managed to bring these two together!
What drew you to volunteer with DataKind UK?
I have been a volunteer with DataKind UK since 2020. I was always amazed by the use of technology for good, especially through software and analytics, which allows one to iterate and scale fast, achieving impact with very few resources.
Through my involvement with grassroots charities and NGOs in the past, I was made aware that even in this day and age cutting-edge technology has not yet penetrated those organisations, preventing them from achieving their true potential. In contrast, larger, for-profit organisations have the resources to attract talent and build rich software ecosystems.
I thought there must be a way to modernise nonprofits and help them make use of their data to improve outputs in a sustainable way. To the best of my knowledge, DataKind is the only organisation that embraces this vision to create impact through data.
What surprised you most about your volunteering experience?
What surprised me was that individuals with different scientific backgrounds came together to work on a project that did not correlate much with their field.
Yet in a few months, we understood the problem and set up a new codebase with preprocessing and modelling facilities, and with decent test coverage. Of course, we also had excellent project management and leadership, but this shows what a team can achieve if they are passionate about a cause.
Is there a resource you’d recommend to the community?
The Deep Learning Specialization on Coursera is a wonderfully laid-out course that really inspired me to delve deeper into data.
The indoor cats, Kremidi (left) and Tamarin (right)
Tell us something completely non-data-related about yourself!
In my free time, I play woodwinds in a band, windsurf, and tend to four cats. Two inside the house, and two strays, which may eventually make it into the house if they behave well.
