About me

I'm Johannes Gontrum,

A software developer with a strong focus on natural language processing, deep learning, and entrepreneurship. My experience includes 4+ years in industry and over six years in research & academia. Currently, I'm working on my early-stage startup textcloud, supported by the Uppsala Innovation Center and a validation grant by Handelsbanken.

I enjoy working as a freelancer to help companies master their NLP and AI challenges. If you're interested in hiring me for consulting, research, or development, please reach out via email.

Get in touch

Name Johannes Gontrum

Language Skills German (native) - English (fluent)

Location Uppsala, Sweden

Interests Entrepeneurship, Language Technology, Polaroid Photography


Education & Skills

Education
2019 - 2020
Linköping University

Freestanding Course:
Ethics in Artificial Intelligence

2017 - 2020
Uppsala University

Master of Arts in
Language Technology

2011 - 2015
University of Potsdam

Bachelor of Science in Computational Linguistics

Experience
2020
Software Development and Consulting - Freelance

Knowing AB

2020
Co-founder

textcloud.co

2016 - 2020
Mid-level Developer for Computational Linguistics

Retresco GmbH

My current stack includes Python (spaCy, PyTorch, Scrapy, FastAPI), JavaScript/TypeScript (Node, Vue), as well as Elasticsearch and MongoDB.

Deep Learning in NLP (PyTorch)

Python

Typescript

DevOps (Docker, Serverless, bare-metal)


My Work

Publications

Gontrum (2019)
Attention Mechanisms for Transition-based Dependency Parsing

Master's Thesis.
Examiner: Prof. Joakim Nivre.
Grade: Passed with Distinction.



Gontrum, Groschwitz, Koller, Teichmann (2017)
Alto: Rapid Prototyping for Parsing and Translation

EACL


Gontrum (2015)
State Splitting for Interpreted Regular Tree Grammars

Bachelors's Thesis.
Examiner: Prof. Alexander Koller.
Grade: Very Good.


Gontrum, Scheffler (2015)
Text-based Geolocation of German Tweets

2nd Workshop on Natural Language Processing for Computer-Mediated Communication


Scheffler, Gontrum, Wegel, Wendler (2014)
Mapping German Tweets to Geographic Regions

KONVENS

Selected Talks

2019
Fine-tuning Contextualized Word Embeddings for Dependency Parsing

Uppsala University.


2017
Building Lightweight APIs with Connexion and Swagger

Retresco GmbH.


2017
Introduction to NLP with spaCy

University of Potsdam.


2015
Text-based Geolocation of German Tweets

2nd Workshop on NLP4CMC.

Selected Articles & Projects

Open-source Project
Dockerized REST API for spaCy

100k+ pulls on Docker Hub, 200+ Stars on Github.


Article
JWT Authentication with FastAPI and AWS Cognito

Blog post published at Data Driven Investor about JWT authentication in the FastAPI framework.


Research Project
Integrating BERT and ELMo into the UUParser

First known combination of contextualized word-embeddings with a transition-based dependency parser.


Research Project
Attention-based Dependency Parser in PyTorch

PyTorch implementation of a transition-based dependency parser. Project for my Master's Thesis.


Team Project
Recommendation Engine for mysimpleshow

Greatly contributed in designing and implementing a recommendation engine while employed as a software developer at Retresco GmbH.


Article
How to build a simple Email List directly from your Landing Page

Blog post that explains how Airtable and AWS Lambda can be used to collect email addresses from your landing page visitors.