grahamw

South Africa, Johannesburg, speaks English, Afrikaans
1210 points (ranked #134)Joined 9 months ago

Programming Languages: Python

Data Science Specializations: Machine learning

What do you enjoy about data science and machine learning?

Data science is more than mathematics: it is the study of meeting human needs through scientific and innovative endeavour. With lateral thinking, research and hard work, I believe that I can bring valuable input to any situation through problem-solving and creative thinking. I have a keen interest in improving the lives of others through science and technology.

Bio

I am a fast learner and grasp concepts quickly and comprehensively. I have always been praised for my ability to detect errors before they turn into problems. I provide solutions that address the issue, rather than hiding the symptom. I thrive on providing valuable insights across a range of fields.

LucidRead

I created a Google Chrome web extension to help people with dyslexia to read on their browser. It works by changing the colours of letters that the user commonly confuses, thus making them easier to distinguish. Check it out at https://chrome.google.com/webstore/detail/lucidreadnnndaacldnholgadkefhdmnnmjnanjjh

Karate Instructor

I am an instructor and assistant instructor at Northcliff Karate Academy.

Awards Received

Awarded 'Most Innovative Team' Altron Award (2018) Awarded Best Masters Presentation in the applied physics track at the South African Institute of Physics conference (2015) Papers accepted to be published in the proceeding of the South African Institute of Physics (2014 and 2015) Awarded a ni-dan (2nd black belt) in Goju-Ryu Karate(2018) Achieved bronze medals at Comrades Marathon (2015-2018)

Oct 2016–present
Netstar
Work
I did an analysis of the effect of a rewards program on driver behaviour. I took bad behaviour alerts from the tracking device, such as harsh braking and acceleration, and used an insurer proprietary algorithm to calculate a monthly score. I then compared the scores across three groups of drivers: Drivers who were unaware that they were being scored, drivers who could see their scores, and drivers who were rewarded, based on their score. I did an analysis of tracking data from buses to identify sabotage, vandalism, and misuse. By doing an anomaly detection on location and network connection, we could figure out when and how the tamper alarms were bypassed.We could then detect battery theft and misuse of vehicles. I created machine-learning algorithms to identify motor vehicle accidents from their tracking data. This informs vehicle insurers that a vehicle has been in a collision, within a few minutes of the incident.
May 2014–Sep 2016
Tellumat Integrated Systems/CK solution
Work
I developed a remote monitoring and temperature control system for use in telecommunication base stations.It monitors a range of sensor- and log information, which can be retrieved remotely. It allows for port forwarding to other devices on the site.
Jan 2014–May 2016
University of Johannesburg, Physics
University
M.Sc. parttime
Apr 2011–Apr 2014
Thate Ubeke Manufacturing
Work
I created the control system for an autonomous land mine-detecting vehicle. Its capabilities include controlling vehicle speed, a tyre inflation system, mine detection system, and lights - for both driver and unmanned situations. I worked on the algorithms used to detect land mines. I created a mission planning package allowing a commander to plot routes, set fields to be cleared, as well as to set caution and keep-out areas for the above-mentioned vehicle on a digital map. These missions are transferred to vehicle-mounted, embedded windows computers, which direct the driver (live or virtual) and log mine detections, locations, and sizes.
Jan 2006–Dec 2009
Rhodes University, Physics and Computer Science
University
B.Sc. and honours
–present
Process Engineer at Malesela Taihan Electric Cable
Work
I created a complete application to design power, copper communications, and fibre optic cables, taking into account strength protection and current requirements, as well as calculating the bill of materials, process requirements, and cost and manufacturing time. This freed up numerous engineering hours, and significantly shortened prototyping time.