Architect becoming a machine learning engineer
How I moved from designing buildings to developing AI-driven applications. An 11-month journey of a 31-year-old architect from Poland.
TL;DR (too long; didn’t read)
- Early days. Since the childhood I had a passion for both science — especially math & computers — as well as arts — music & handcraft.
- Education. I decided to pursue my education in architecture and got my degrees in Poland & UK. Nevertheless, I maintained strong interest both in music as well as computer science.
- Post-univerisity. After graduation I worked as an architect in Norway, Germany & Belgium. At the same time I was boosting my skills as a 2d & 3d graphic designer.
- JEJU.studio. Eventually I moved back to Poland and co-founded an architecture office and worked on number of projects including a school based in the former refugee settlement in Tanzania.
- Artificial Intelligence. Driven by an interest in the technological development as well as a profound need to contribute to the better future of the world I got fascinated by Artificial Intelligence.
- Learning path. I started learning machine learning & data science concepts as well as Python coding with the help of various online courses.
- Personal portfolio. A number of ideas for personal projects pushed me to bring my theoretical knowledge into practice and build my personal portfolio.
- First job. After 10 months since I started learning I landed myself a Data Science traineeship which got my to the Junior Data Scientist position in two months.
Early days
I have always been fascinated by two things: science 🔬 & arts 🎨. I’ve loved maths since primary school. I felt enchanted by an underlying beauty of an abstract, yet logical and intertwined language of mathematics. It gave me tools and motivation to explore the world of science and technology, especially computer-science 💻.
May not music be described as the mathematics of the sense, mathematics as music of the reason? The musician feels mathematics, the mathematician thinks music: music the dream, mathematics the working life.
This quote by James Joseph Sylvester points to another passion of mine — music 🎵. I was raised in a home where arts played a significant role. My father used to paint, my mom & two sisters each played instruments, so did I. Besides playing a piano and a mandolin I was also keen on hand-drawing ✏️.
Education
As a teenager I started designing and coding my first hobbyist websites👨💻️. In parallel, my handcraft skill which eventually got me interested in architecture. When it came to deciding the field of study I was hesitant between computer-science and architecture. Both of them seemed exciting but the one which felt more real and tangible was the final pick — architecture🏙️.
Even though the decision had to be made I didn’t give up on computer-science. I kept on working on some personal web-development & web-design projects which at some point evolved into part-time work. The jobs helped me boost my budget 💰 a lot, but also stimulated my interest in the field.
During my university years I further developed my passion for music and co-founded a music-discovery platform called eargasm music 👂. Currently the platform is moved to a Spotify channel and hosts number of curated, thematic playlists . Feel free to head over and find if there’s something eargasmic for you! 🎧
Post-university
After graduating from the Faculty of Architecture at Wrocław University of Science and Technology 🇵🇱 & Lincoln School of Architecture 🇬🇧 I started my career as an architect working for renowned architecture offices all around Europe including Saunders Architecture 🇳🇴, Heupel Architekten 🇩🇪 & OYO Architects 🇧🇪.
In the meantime, I expanded my existing computer-science skill set with some new abilities. I boosted my proficiency in 2d- & 3d-graphic design 🖼️, to the point where I considered establishing a 3d visualization practice. Eventually, instead I decided to move back to Poland 🇵🇱 and co-founded an architecture office JEJU.studio together with my good friend — Iwo Borkowicz.
JEJU.studio
Together with Iwo we’ve had number of opportunities to work on some exciting projects of different scales. From a commercial set-design & scenography 🎭, through two over 5-hectare large mixed-use developements 🏗️, to a primary school🎓 in a former refugee settlement in the countryside of Tanzania 🇹🇿.
The Ulyanulu Pre- & Primary School we designed and built is the one I’m the most proud of. I strongly believe in the urgent need to tackle two biggest problems of today’s world: inequality👦🏻👨🏿 and climate change🌍 and this project helped me to make my contribution to the better future of the world.
Artificial Intelligence
Nevertheless, a questionable agency😩 of architecture community, the closed-source🚪 nature of the field, an occupational burnout🔥 and a disruptive nature of a global COVID-19 pandemic🦠 got me wondering. I decided I need some fresh air.
I wanted something open-source, something proactive, something that can give me an ability to bring an added value out of thin air without resource- and time-consuming processes. Back to 3d graphics? Web-development? Nah… I needed to make a bold move to make me wake up in the morning excited yet again! 😃
I remembered a great article by Tim Urban’s Wait But Why I stumbled upon a while earlier — The AI Revolution: Our Immortality or Extinction. It clicked! Artificial Intelligence and its underlying machine learning techniques were going to be my daily driver since then. I was about to start discovering behind-the-scenes of the jaw-dropping accomplishments of Elon Musk (OpenAI, Tesla, Neuralink); mind-boggling ideas of Ben Goertzel (singularityNET); inspiring narratives of Lex Fridman (YouTube channel). Intimidating, but incredibly hair-raising… 🤩
Learning path
Where do I start?
I immediately started looking for a place to start learning. I asked around people I knew who might have known something about the field. 💡💡💡 I remember talking to a brother of my best friend who told me about a free machine learning course by Andrew Ng, whom he would describe as a male version of Mrs Purzycka 👩🏫, our secondary school maths teacher, who is one of the best educators I’ve known, and who is the important reason why I love mathematics in the first place. This course seemed like a perfect starting-point.
ML by Andrew Ng
Even though I still had some ongoing responsibilities with my architecture project I jumped right into the course. It’s a series of reading materials and video lectures featuring Andrew himself explaining everything in a really laid-back style 😎. The materials are mostly theoretical but fairly easy do digest by anyone with some basic computer science knowledge. It helps to develop a solid intuition how machine learning algorithms works and what it can be used for.
Each concept/chapter is concluded by graded review questions and coding exercises. They require you to complete missing parts of the source code which put the theory from the lectures into Matlab code. Even though none of the exercises is particularly exciting, they give you an deeper understanding of what’s happening under-the-hood of the libraries and frameworks you’ll use later on 🕵️. You can find the course at Coursera. I managed to finish the course and then redo the exercises to revise the knowledge within two months doing it only part time.
Maths
There’s quite a significant amount of mathematics involved in the field — linear algebra, calculus, probability & statistics 🔢. I studied most of the concept already during my high-school as well as university, however I found it really helpful to revise some of the knowledge. There’s plenty of great resources online for both learning those concept from scratch or just revising your knowledge like Seeing Theory by Daniel Kunin or the amazing Kahn Academy 🤯, but there’s plenty of other resources on Youtube ▶️ and Medium ✍️.
Python & SQL
The Coursera’s ML course used Matlab as a programming language, however the most popular language used in machine learning is Python - that’s why it became my language of choice 👨💻️. I didn’t really have a strong coding background — I only had previous experience with web development (HTML, CSS & basics of JS & PHP) and Bash.
To grasp some basics of Python & SQL I stayed at Coursera platform as I found it convenient to use and jumped into Python for Everybody by University of Michigan tutored by a charismatic tutor — Charles Russell Severance. The specialization consists of 5 courses covering various topics. On top of the basics of the language such as its functions, data structures etc. the specialization also covers regular expressions, accessing web data, using databases as well as retrieving, processing, and visualizing data.
The specialization is really straightforward, well-explained and really easy to go through, so within a week I learned enough basics to put them into work and learn new thing along the way from Python Documentation.
Other courses
After getting comfortable with Python and the basic machine learning concepts I dived deeper into learning. I picked two courses: the first one was Applied Data Science with Python by University of Michigan and the other one Deep Learning by deeplearning.ai. Both of them are 5-course long specializations and cover a wide range of topics.
Personal portfolio
Even though the courses I finished were full of knowledge I found it way more efficient to learn while doing stuff rather than following video lectures and filling gaps in the exercise code. Therefore I decided to jump into some personal projects right away.
Don’t worry if you get stuck on something — most probably you’re not the first one encountering a certain problem 🤦. There’s plenty of good resources on StackOverflow forums, GitHub repositories and of course documentations. They’ll be your best friends! Just make sure not to blindly copy-paste the code snippets. Try to understand them, rewrite them, experiment with them and always comment them, so you can always come back to revise the knowledge 🙇♂️!
Projects
When looking for inspirations for my personal project I would always look for topics which are of my interest to keep myself extra-motivated. As I mentioned before I’m a huge fan of music 🎧. That’s why for my first project I decided to tackle the music library of my Eargasm music blog and building a playlist classification algorithm which was meant to predict which playlist should new songs be assigned to.
After finishing it I tried to complement my original profession — architecture — with artificial intelligence and create an object and color detection search engine for one of the most popular architecture websites Dezeen 🔎.
Another idea of mine was to to create a project a broader audience might take advantage of. I created a web-scraper which collects the information about the deputies of the current term of the Polish Parliament 🏛️. The goal was to build a comprehensive exploratory data analysis tool for a greater transparency of the Polish authorities 🕵.
Apart from that I also did some smaller projects to practice my EDA & classification skills on the well-known Kaggle datasets such as Red & White Wine Quality 🍷. I will gradually go about writing more in-depth articles about the projects I worked on, so stay tuned.
First job
Ten months of hard work did finally bear fruit and after a recruitment process I landed myself a Data Science traineeship at business intelligence company in my city, which after two months became a Junior Data Science position. To be continued…
