Hey! My name is Tobias Cornille
& I'm a creative software engineer

JavaScript TypeScript React React Native Python Three.js Vue PyTorch Hugging Face Django AWS

Selected projects

Multi-Sensor Data Labeling Platform

Segments.ai


Segments.ai is a YC-backed startup building a data labeling platform for robotics and AV. I joined as the first engineer, and created Segments.ai's 3D labeling suite. I also worked on ML-powered image segmentation tools based on Transformers and SAM.

  • A full-featured labeling interface for 3D and multi-sensor data, built with Three.js and Vue

  • Created Autosegment, an AI segmentation tool, and Panoptic-SAM, a pipeline for using Segment Anything for panoptic segmentation

  • Developed and deployed a point cloud processing service using Docker, AWS Step Functions, AWS Lambda, and AWS Batch

  • Wrote several blog posts on computer vision, machine learning, and 3D web development, some of which were featured on Hugging Face, Weights & Biases, and codrops

  • Redesigned the web platform's UI with a focus on simplicity


Vue Three.js Python Django Hugging Face AWS

Expressive Speech Synthesis

KU Leuven - Acapela Group


For my Master's thesis, I developed ConEx, a text-to-speech model that can produce speech in a certain speaking style, while also allowing local prosody adjustments. ConEx builds on a powerful Transformer-based TTS architecture, but adds a reference encoder to learn global prosody embeddings, and a VQ-VAE to create fine-grained prosody embeddings. I also proposed a new method to edit prosody locally. After training the model, I showed that the model enables multi-level prosody control by performing experiments on two datasets.The thesis received a grade of 19/20. It also made it to the finals of the EOS prize, the prize for the best scientific thesis in Flanders. Furthermore, the thesis was the basis of a paper titled Interactive Multi-Level Prosody Control for Expressive Speech Synthesis, which I wrote in collaboration with my thesis supervisors. It was published at ICASSP 2022, the leading conference on acoustics, speech and signal processing.


PyTorch ESPNet

Biomedical Data & NLP

Invitae


In 2020, I worked at Invitae as a full-stack intern. More specifically, I worked on parts of Moon, which is an AI tool that diagnoses rare diseases from genetic sequence data.My first project involved extracting data from biomedical databases, transforming the data, and then extending a part of the UI to visualize the data.Secondly, I improved a text labeling algorithm that helps employees process academic papers more quickly. I ran experiments and implemented an improved machine learning algorithm in Python using the spaCy and Hugging Face. At the end of my internship, we successfully deployed the improved service.


spaCy pandas Ruby on Rails Hugging Face

Cross-Platform App for The Telegraph

Twipe Digital Publishing


In 2019, I worked as a front-end intern at Twipe, a company developing digital solutions for leading newspaper publishers. During my internship, I worked on a new app for the Telegraph written in React Native. The mobile engineering for this project consisted of one colleague and myself. I thus had a lot of responsibility as an intern and got to write production code from day one. This allowed me to learn and grow quickly as a professional developer.


React Native Redux TypeScript

Mobile Browser-based Game

KU Leuven


Game of Wolves is a browser-based game similar to Pokemon Go. It uses technologies like face recognition and augmented reality to create an exciting user experience.We developed this game with a team of 6 students during the 3rd year of the bachelor of computer science. I was responsible for developing the front end using React.


React.js Redux node.js mongodb


Education

MSc Computer Science
KU Leuven

Major in Artificial Intelligence
Graduated Summa Cum Laude

2019 - 2021

Erasmus+ Exchange
Técnico Lisboa



2019 - 2020

BSc Informatica
KU Leuven

Graduated Magna Cum Laude

2016 - 2019

Professional Experience

Full-Stack Engineer
Segments.ai

Joined as the first engineer and created Segments.ai's 3D labeling suite. I also worked on ML-powered image segmentation tools.

2021 - present

Full-Stack Engineer Intern
Invitae

Implemented a machine learning tool to automatically categorize abstracts & extended their web app with biomedical data.

2020

Front-End Engineer Intern
Twipe

Internship at a digital publishing company: worked on a cross-platform newspaper app for The Telegraph.

2019

Volunteering

Hackathon Organizer
Academics For Technology

Main organizer of WhatTheHack, the biggest student hackathon of Belgium (~250 participants).

2017 - 2019

Marketing
Academics For Technology

Implemented a strong marketing strategy. Designed marketing material and created a style guide.

2020 - 2021

Member
BEST Lisbon

Member of a European student organisation, helping with graphic design and event organization.

2019 - 2020