profile picture

Dominik Hollidt

Master Student Computer Science at ETH

Languages

  • German (Native)
  • English (Fluent)
  • Swedish (Beginner)
  • Latin ()
  • Spanish (Beginner)

Interests

  • Climbing/Bouldering
  • Mountain Biking
  • Badminton
  • Playing the Accordeon and Piano

About Me

I am Dominik a computer science student at ETH. I am always looking for opportunities to learn more and jump out of my comfortzone. My main focus is on machine learning. However, I try to gain knowledge in many areas. I could demonstrate interdisciplinary knowledge as a team member of two formula student teams. There I could not only gain hands-on machine learning experience, but also classical algorithm design and team management with the goal uniting software and hardware needs.

Publications

  • Geometry Aware Field-to-field Transformations for 3D Semantic Segmentation
    • Dominik Hollidt, Clinton Wang, Polina Golland, Marc Pollefeys

    Education

    Master Thesis

    March 2023 - August 2023
    MIT (CSAIL, Medical Vision Group - Prof. Polina Golland) - Cambrige, USA

    Worked on deep 3D semantic segmentation from 2D supervision only. Utilized NeRFs and field-to-field transformation via point cloud transformers. Investigated the effects of pretraining on fine tuning performance of 3D semantic segmentation.

    Master Studies

    September 2020 - now
    ETH Zürich - Zürich, CH

    Excited about more international experience together with the chance to study at a top 4 university for computer science I started my master studies at ETH Zürich. With the major in Machine Intelligence and minor in Theoretical Computer Science to improve my algorithmic understandings.

    • Machine Intelligence
    • Theoretical Computer Science

    Erasmus Exchange Student

    August 2019 - June 2020
    KTH Royal Institute of Technology - Stockholm, SE

    To learn more about other cultures, learning experiences at other universities, improve my language and grow as person I went on exchange to KTH. Here I developed a stronger interest in computer vision and wrote my Bachelor Thesis about Optical Flow Algorithms for Event Based Cameras

    • Machine Learning
    • Computer Vision
    • Robotics
    • AI

    Undergraduate

    September 2017 - June 2020
    Technical University Berlin - Berlin, GER

    Started my degree at TUB. Finished the basic studies plus business class to extend my horizon. Graduated with 1.3 German Score.

    • Python
    • Java
    • c++
    • basic studies

    Pre-University Classes

    Summer 2017
    Free University of Berlin - Berlin, GER

    Excited to start my degree I spent the summer at FU Berlin with introduction classes to computer science. This confirmed my motivation to study computer science.

    Student Graduation

    July 2017
    Eckener Gymnasium - Berlin, GER

    I graduated from school in 2017 with 1.4 German Score. Already in school I was excited to extend my horizon and had an exchange year to Thomas Adams School in Wem, Great Britain. Focus and elective classes: Chemistry, Math, Computer Science

    • Chemistry
    • Math
    • Computer Science

    Projects and Internships

    A list of internships and smaller projects I have worked on in my free time:

    Daedalean - Worked in the Spotter Visual Traffic Detection Team. My main focus was on the generation of synthetic flight scenarious from real historical flight data to evaluate the intruder detection accuracy, togheter with a statistical analysis of historic flight data. Further, I worked on reducing the domain shift of simulated flight scenarios by automatic color grading. Further, I worked on an automated upload station to upload large amount of flight test data to the cloud. Lastly, I worked on the whole data evaluation pipeline and toolbox.
    Formula Student - Formula Student is a student engineering competition with a huge community around the globe. The goal is to build an autonomous race car and compete against other university teams around the world, e.g in Hungary (FS East), Italy (FS ATA), Germany (FSG) and many more. Here hundrets of student teams come together and demonstrate their technological progress and developments. This is not only happening in autonomous race cars but also in combustion or electric cars with drivers.
    KTH Formula Student (08.2019 - 06.2020) - In the first year in the formula student world I learned about machine perception with the focus on LiDAR. Further, did I help with team management (12 man team) and experimented with learning in the cloud with Docker.
    AMZ (09.2020 - 09.2021) - Stepping up the game working almost full time I joined the Formula Student Team at ETH. Here I developed a LiDAR color detection method via a CNN. Further, I worked a lot with Software Infrastructure, Continuous Integration and a custom KPI tool (Jenkins, Docker). Most valuable was to learn about the whole end-to-end integration of our autonomous pipeline together with the software-hardware integration (developed automatic LiDAR calibration tool).
    Internship Data Analysis AVM (06.2020-09.2020) - The goal was to perform data analysis of customer support data and automatically link it to crashes and bugs. Developed automated generated data summaries via Spark and Zeppelin, a tool to analyse Wifi speed tests and a tool to see feature relevance via decision trees.
    Wiki-Alt Text Tool - Images on the internet have an alt text which describes the image. This is important for people with visual impairments. I worked on a tool which fetches random wikipedia images, lets the user annotate the alt text and submits the updated alt text as a change request to the wikipedia page. This way the wikipedia page can be improved and the user can help people with visual impairments.
    Weather Transfer via GANs - During the course Deep Learning I worked on a project to transfer the weather of an image to another weather. The goal was to decrease the uneven distribution of data in different weathers.
    Drug Discovery via NLP and Transformers - Used Transformers on knowledge graphs to predict new drug candidates. The goal was to predict new drug candidates for a specific disease. The knowledge graph was build from a database of drugs and their properties. The model was trained on a subset of the knowledge graph and then used to predict new drug candidates.
    AR Memory Game - Developed an AR Memory Game for Android via Unity and Vuforia. Visually impaired people can not play the normal memory game. Hence, we used other modalities to make the game playable for them. The phone vibrates gives an audio signal when the user hovers over a card.
    Free Text Search in Text - Searching Text in PDFs requires an exact match. I developed a small tool which can take in a text and it can be searched via free text queries and questions using pretrained Language models.
    Remote Control LED Stribe - Do you know these cheap LED Stribes which can light in all colors? What do you do if the battery of the remote control runs out? Correct you connect it to a microcontroller and build a small app to change the colors. To have even more fun you make it react to music. Now you can have a party in your house with all LED Stribes being controlled from one source.
    Cat Feeding Automaton - A machine to feed your cats. That would make your life so relaxed. Thats what I thought as well hence I build a small machine intended to feed cats automatically. It will dispose portions perfectly for your cats without you doing anything.

    Awards and Scholarships

    Honorings and scholarships I received:

  • Spickenreuther Foundation Scholarship for Master's Thesis at MIT
  • Swiss-European Mobility Programme for Master's Thesis at MIT
  • VDI Honoring - For excellent graduation
  • ERASMUS Scholarship - Exchange to KTH
    • My Exchange to KTH, Stockholm, was supported by the ERASMUS+ Scholarship.
  • IT-Talents Scholarship
    • As part of the IT Talent Scholarship Dok Systeme supported me financially.
  • Nominated for "Stipendium des deutschen Volkes"
    • Due to excellent academic perfomance I got nominated for the "Stipendium des deutschen Vokes".

    Skills & Proficiency

    C++

    Python

    ROS

    Keras

    PyTorch

    Numpy/Pandas

    Docker

    Matplotlib/Plotly

    Jenkins/CI

    Java

    (py)Spark

    Ruby

    German (native)

    English (C1)

    Swedish (B1)