As a university student in my third year, I had the opportunity to assist in teaching first year students Python. Being a teaching assistant reinforced my knowledge, by encouraging active recall of the modules I'd studied. The best part about teaching is constant learning, whether that's learning to better communicate a concept, or learning something completely new in order to be able to teach it.
I'm currently learning French, which for me has been a major learning curve as I've never learnt another language before. Teaching English to French speakers (intermediate/near advanced) has really helped me to think about English language structure and how it compares to French.
I've also lead a volunteering project that teaches the elderly how to use computers, mobile phones and tablets. Although the level of complexity was in stark contrast to teaching programming, it still offered me challenges such as further simplifying an already 'simple' concept. Encounters like these kindled my creativity, since I had to think outside of the box that I was so used to. Volunteering offered me a chance to give something back to the community and it was rewarding to see how far those efforts went!
Microsoft Open Hack
Whilst interning at Cubic the opportunity to attend this machine learning hackathon came up. On arrival I was shown my group (18 teams of roughly 100 delegates from a number of different companies and organisations).
We faced a series of seven structured challenges to solve problems in the Computer Vision space
Challenge 0: Setting up group environments (creating clusters, DataBricks or VMs, Jupyter Notebook)
Challenge 1: Training using prepackaged Azure cognitive services
Challenge 2: Data wrangling
Challenge 3: Classical Machine Learning Algorithms
Challenge 4: Deep Learning (Convolutional Neural Networks)
Challenge 5: Containerisation
Challenge 6: Object Detection
One of the highlights was classifying and determining objects. This was preceded by normalising images to ensure they were ready for processing (if you feed junk into your algorithm ...you'll get junk back out). Once the data wrangling was complete I was able to see the varying difference between how the classical machine learning algorithms performed on the given data set, when compared to deep learning. For image classification the deep learning algorithms performed best. Sep 18
Team 16, the four of us and our proctor
Winchester Council held a hackathon at the University of Winchester. One challenge was to hack together solutions that help improve local business using machine learning algorithms to process data. Our solution was the 'Staff optimiser' which was to be aimed at local stores. It's a simple prototype that uses a trivial linear regression algorithm but still managed to demonstrate the overall concept (to the demands of this hackahon).
A screen is displayed at checkout asking for a score.
The figures are stored and tracked over time.
Businesses can determine where staff work best. Perhaps have the staff with highest customer feedback on the tills and lowest in the stock room. Or maybe change shift patterns if one person scores better in the morning and another in the afternoon. Sep 2018
Reading University: R.U Hacking?
1. Being unable to allow access to someone until you arrive home.
2. Losing your house keys.
The solution: A remote/local door unlocking device we called 'FaceLock'.
Made of three parts: user web interface, a server and a facial recognition device.
I took responsibility of the facial recognition aspect which involved programming a Raspberry Pi in Python using the OpenCV library as well as implementing the electronics aspect (camera, buttons, motor and LED connectivity). This was really my first introduction to OPenCV, but a useful one.
The device would first detect a face (based on its trained understanding of what a face is) and then capture and image.
The image was compared to what was stored on the database and if matched, would turn the motor (simulating a door lock).
We created several prototypes that approached the project in a different way. In the end we were able to demonstrate a working project although not the way we would have desired as it didn't entail the full functionality we envisioned.
However, given the 24 hour time constraint it was still something achieved, with judges awarding us second place. Jan 2018
Collecting our awards
Few instruments have anything close to the range of the piano – both tonally and dynamically.
At times it's relaxing, at others it's a mental assault course. Either way, playing sharpens concentration, teaches perseverance and instills discipline
Once time's invested, the joy from producing pleasurable music is one of the best things on earth! Of course, we all have a varying understanding of 'pleasurable music', but a love of music is virtually universal.
I play both badminton and golf. Badminton offers great teamwork opportunities as communication with your teammate is crucial. I find golf, on the other-hand, more collective and it has proved useful in maintaining composure.