Korbinian Koch (he/him)

Korbinian Koch (he/him)

MSc Intelligent Adaptive Systems

University of Toronto

University of Hamburg

Hey! 🤗

I am currently pursuing a Master’s degree in Intelligent Adaptive Systems at the University of Hamburg, which I will finish in 2023. In the past, I have interned at the Vector Institute for Artificial Intelligence in Toronto and MOIA, a ridesharing startup based in Hamburg.

Download my CV.

Interests
  • Artificial Intelligence
  • Representation Learning/Embeddings
  • Fairness in ML
Education

  • MSc Intelligent Adaptive Systems, ongoing

    University of Hamburg

  • Graduate Exchange Student, 2022

    University of Toronto

  • BSc Human-Computer Interaction, 2020

    University of Hamburg

  • Undergraduate Exchange Student, 2018

    University of Calgary

Experience

 
 
 
 
 
Vector Institute
Applied Machine Learning Intern (part-time)
Vector Institute
Jan 2022 – Apr 2022 Toronto, Canada
  • Developed a customizable reranking framework for Reddit threads, increasing the diversity of viewpoints shown to users using community- and language-model-based scores.
  • Built a user study interface using GCP and Streamlit for the evaluation of said scores.
  • Wrote a Slack notification tool that tracks the status of long-running experiments on own initiative in second week.
  • Ensured participant success as technical facilitator in Vectors forecasting bootcamp, a 3-day workshop introcuding participants to state-of-the-art ML forecasting techniques.
  • Improved the documentation of NeuralProphet, an open-source forecasting framework.
 
 
 
 
 
MOIA
Cloud Native Development Intern (part-time)
MOIA
Nov 2019 – Apr 2020 Hamburg, Germany
  • Developed microservices in Kotlin using REST, GraphQL and a variety of AWS services such as Lambda, RDS and SQS.
  • Optimized data quality and query performance on PostgreSQL database.
  • Convinced management to allow students and interns to take part in team self-selection.
 
 
 
 
 
University of Hamburg
Teaching Assistant
University of Hamburg
Oct 2017 – Mar 2018 Hamburg, Germany
Teaching assistant for the course Software Development I in Java. Responsibilities included supervision of classes of ~30 students, giving feedback on and grading assignments.

Projects

Dec 15, 2021
DisCeRn: Disease-Contrastive Representations
DisCeRn: Disease-Contrastive Representations

Improving chest x-ray representations by contrastive pre-training on only marginally related pathologies with advanced negative sampling and loss calculation.
Dec 12, 2021
Unlearning Fairness
Unlearning Fairness

An investigation into the effects of different unlearning strategies using SISA on the fairness of models for protected groups.
Aug 24, 2021
Modergator: snapping back at hate speech
Modergator: snapping back at hate speech

A Telegram bot that moderates groups for hateful texts, voice messages and memes using a selection of pre- and self-trained models.
Jul 30, 2021
Dreaming in goal-conditioned environments
Dreaming in goal-conditioned environments

An adaptation that makes the reinforcement learning agent Dreamer work in goal-conditioned OpenAI Gym environments.
Oct 1, 2020
Predicting speaker quality using embeddings
Predicting speaker quality using embeddings

My Bachelor’s thesis on the usage of speech embeddings pre-trained for speaker verification for the prediction of perceived speech likability.
Sep 1, 2020
Predicting droughts using meteorological data
Predicting droughts using meteorological data

A small experiment on the effectiveness of random convolutional kernels (ROCKET) for drought prediction.
Mar 25, 2020
A friendly introduction to text clustering
A friendly introduction to text clustering

A Medium article covering word and document clustering using k-means, brown clustering, tf-idf, topic models and latent Dirichlet allocation. 60k+ views, featured by the editors.