ABOUT ME
Hello there!
My name is Timothy Fiss and I am a Junior Computer Developer from
Regina, Saskatchewan. I have a passion for technology and a deep
interest in the field of computer science. I am constantly seeking
opportunities to expand my knowledge and skills in the ever-evolving
world of software development.
With a strong foundation in programming languages such as Python,
Javascript, and C++, I am confident in my ability to develop
high-quality software applications. I am also familiar with various
technologies and tools including Agile methodologies, Git, and
SQL.
In a previous role I played a key role in enabling cross-team
communication and providing technical support. I am a quick learner
and always willing to go the extra mile to ensure projects are
completed to the best of my ability.
Aside from my technical skills, I am a team player and value strong
communication and collaboration in the workplace. I am always
looking for ways to contribute to the success of a project and
actively seek out feedback to continuously improve my
performance.
I am excited about the opportunity to grow my career in the tech
industry and take on new challenges. I am open to new learning
opportunities and am eager to continue to expand my skill set.
Projects
Optical Flow
In 2017 I read a
paper
on using the optical flow algorithm to measure the behaviour of
poultry, correlating it to several welfare measures in poultry.
optical flow is a method commonly used in computer vision to measure
the movement of pixels in a video stream. The author of the paper used
specialized equipment to collect the video of the poultry, and focused
on real-time analysis of the birds. This made it ill suited for the
research environment at the University of Saskatchewan where I was
studying at the time. At the U of S we already had infrastructure
installed to allow for the collection of video which was incompatible
with her system. I had a desire to replicate her work in a way that
would accept any standard video input, allowing for post-hoc
evaluation of the metric. At that time, I had no formal computer
science experience or education, so this project exceeded my
capabilities. I was successful in creating a function that could take
a video input and perform the analysis on it, but it was not efficient
or performance-optimized.
In 2020 I revisited the project to demonstrate to my colleagues
at Chicken Farmers of Ontario how it worked and discuss its
applications in the poultry industry. At that time I modified it to
create a simple live demonstration.
You can check out the project on my GitHub
here
Fraction Estimation
In Early 2021 a friend of mine wanted to know the fractional grade
they got when they recieved a grade a 76.83% (They only had a rough
idea of how much the project was worth). This seemed like a problem
that a computer would be well suited to solving.
The first solution I explored was using continued fractions to get
closer and closer to estimating the desired decimal. This was fast,
but its output was sparse, in this case missing 7/9 which was the
simplified version of the mark he recieved: 35/49.
The next solution I explored used the Farey Sequence to produce the
results. I found an algorithm that would produce the next fraction
between two fractions in the sequence. So the fractions 1/3 and 1/2
are next to each other when the denominators are less than 4. The
algorithm with these inputs would produce 2/5 as its the next fraction
between the two. My program used these fractions to output
increasingly accurate (fully reduced) fractions until the exact
representation was determined. While not as efficient as continued
fractions, this algorithm would produce a complete list of fractional
esimations starting from the lowest denominator, going to the largest.
It was limited by float accuracy, but it only accepted values between
0 and 1 where floats are their most accurate to limit this inaccuracy.
This programs performance is held back by my homemade greatest common
divisor algorithm. Faster execution could likely be achieved if
performance was the highest priority.
You can check out the project on my GitHub
here
Particle Emitter
Done for a class on animations, this particular program was
laser-focused on producing as many particles using the webgl framework
as possible. It achieves this by performing as much of the computation
on the graphics card as possible. On my 2080Ti I experienced
significant performance bottlenecks at around 80,000 active particles.
While this program is already quite effienct, there is still
significant room for improvement. My understanding of linear algebra
as it applies to graphics was still primitive and there are several
optimizations available. I could also shift some of the workload back
to the js to help balance the program better as the js currently
represents only a small portion of the total frame time. There is also
an opportunity to create a fixed object buffer that would have reduced
the amount of memory reallocation that the js had to perform each
from, reducing the likelyhood of a long frame due to garbage
collection.
you can view the output
here
You can check out the project on my GitHub
here