Jake Pencharz
An engineer who never wants to stop learning.
General Info
- Date of Birth8 Dec, 1994
- AddressFuggerstr. 1
- Munich, Germany
- E-mailjake.jay.p@gmail.com
- Phone (+49) 1520 6247563
An engineer who never wants to stop learning.
A simple implementation of data hiding. Exploiting the fact that no visible information is contained in a pixel's least bit (LSB), these bits are substituted with other information. This technique allowed us to embed the entire first chapter of The Fellowship of the Ring into one image.
My final year project at UCT. Much of the South African deaf community are illeterate. To afford this community equal rights, they should be communicated with in their language of choice, South African Sign Language (SASL). A tool that can translate from English to SASL would help bridge the communication gap between the deaf and hearing communities in South Africs.
An extension of NeRF (Neural Radiance Fields) to apply 3D style transfer using pytorch (pytorch-lightning). The Neural Style algorithm is based on the loss network proposed by Leon A. Gatys, Alexander S. Ecker and Matthias Bethge. Our implementation takes advantage of the way that NeRF isolates the scene's density in a 'head' of the model which is seperate from the 'head' of the model that generates the view dependent color.
Several model are explored to determine the relationship between dopaminergic neurons and mushroom body output neurons (MBONs) in the fruit fly mushroom body. This brain area is essential for forming associative memories related to odors and could shed light on memory mechanims in general. Image from Li et al.
A glove with an accelerometer mounted on the back used to help diagnose Parkinsons disease. Many patients who suffer from this disease tend to tremmor at a recognisable frequency. By analysing the signal generated by the glove in real time, this marker frequency can be checked for.
This Watsapp group used the Google Translate API to allow every member in a watsapp group to commincate in their language of choice. This was a one day project made in collaboration with the team at Kandua.
Working as a research engineer in Bayer's Machine Learning Research Group I have contributed to the development of (soon to be) open source Python tools, and semantic segmentation models aimed at Digital Pathology use cases. The project falls under a broader explainable AI program and by working closely with expert pathologists, aims to deliver and visualise the model's predictions in the most useful format possible.
Working in the Computation in Neural Circuits Group supervised by Shuai Shao and Julijana Gjorgjieva. I worked on modelling the relationship between dopaminergic neurons and Mushroom body output neurons in the Drosophila mushroom body. I built and compared linear models and a recurrent model of the circuit. The results were compared to recently published connectomics data.
Wrote tutorials for a course on Brain Computer Interfaces using EEG (g.tec Unicorn Hybrid Black). The tutorials included using the Unicorn C API to log data to the lab streaming layer (C++), filtering techniques, visualisation, and a primer on motor imagery. These tutorials were used as a crash course for students competing in the BR41N.IO Hackathon (where I competed with my team as well).
Isazi is a small but prominent company in Johannesburg that specialises in machine learning and optimisation. I was tasked with building a data curation pipeline to generate, tag, and manage new training datasets that would be used to tune their optical character recognition (OCR) system. While there, I: Designed and built a PostgreSQL database to ingest and store image data; built a RESTful API with Python (Flask) and a frontend with Vue.js to interact with the database; generated synthetic handwriting data; retrained image classification models on curated datasets boosting performance by ≈10 %; worked with a multidisciplinary team of data engineers, mathematicians, and software engineers.
I spent eight weeks at Isazi working as a an intern software developer. I worked on two projects while I was there. The first was an Android Application that would be used as a data labelling tool. The second project involved developing a website, also for data labelling, using Python on the backend and Vue.js for the frontend.
While studying at WITS, I worked for The Lab Maths and Sciences Academy tutoring high school students. I found the experience extremely fulfilling since it forced me to understand fundamental concepts from many different angles.
The program is interdisciplinary and combines experimental and theoretical neuroscience with profound training in engineering. I have focused my electives around building a strong mathematical grounding and exploring deep learning with some focus on computer vision.
At UCT I gained skills as an electrical engineer, while also being taught the basics of computer science. On the engienering side, the courses covered analog and digital electronics, as well as embedded systems, control engineering, and signal processing. The course that I enjoyed most was probably control engineering. It required me to apply my mathematical skills as well as providing the oppertunity to practice circuit design and some embedded software design. The computer science department covered material from the fields of database design and management, computer architecture, and parallel computing. From this array of courses I found the parrelel computing to be the most challenging and stimulating. I thoroughly enjoyed the balance of all these subjects.
I maintained a GPA of over 80 % for the duration of the degree.
Graduated with distinction. This degree is a combination of Electrical Engineering and Medicine. I was priveledged enough to learn anatomy, via weekly dissections of a cadaver, while simultaneously studing courses like embedded system design and signal processing.
I am very interested in applying cutting edge technology to the medical field. I think that this is a powerful avenue for achieving a high impact in the healthcare sector. This degree was an excellent starting point, and I would jump at any oppertunity to further my developement in this realm.
I graduated with distinction from this course, and maintained an aggregate GPA of over 80 % for its duration.
Graduated with 10 distinctions, including Advanced Programme Mathematics, and Visual Arts. I was also actively involved in school sports, and the public speaking team.
Android Studio
Firebase
Arduino
Matlab
Bash
Git
PostreSQL
