Jovan Cicvarić

Professional Experience

Optocycle GmbH, Tübingen

February 2024 - present | Machine Learning Engineer

Developed classification model for different kinds of construction waste, achieving 90+% accuracy
Developed model for estimating object sizes with monocular camera
Developed labeling pipeline and supervised labeling of more than 20k image across multiple teams
Developed training tracking and model deployment pipeline using MLFlow and Triton
Developed model for plastics classification by combining multispectral classification with LLM
Conducted studies exploring classification performance in multispectral vs RGB domains
Tech stack: Python, Pytorch, NumPy, OpenCV, Docker, Kubernetes, RabbitMQ, CVAT, Triton Inference, MLFlow, Helm, Poetry, Multi/Hyper-spectral Images

Autonomous Vision Group, Tübingen

June 2023 – present | Research Engineer

June 2021 – May. 2023 | Research Assistant

Full stack development of a scientific paper recommendation system Scholar Inbox
Using personalized models trained on users’ previous votes and selected authors for making recommendations.
Developed a registration pipeline that facilitated over 20k+ new users
Optimized SQL queries, Python code, and our embedding storage system for lower latency and memory usage extensively.
Creating a dataset of abstract annotations, containing 2425 sub-sentence labels within 691 annotated abstracts.
Tech stack: Flask, React, HTML/CSS, SQLite, PostgreSQL, Python (Pandas, NumPy, Celery, PyTorch), NLP

Moscow Institute of Physics and Technology, Moscow

December 2019 - April 2020 | Student Assistant

Assisted in preparatory university courses in Mathematics and Physics for 10th-grade students.
Checked homework, gave personal reviews and personalized study guidance.

Education

University of Tübingen

M.Sc. Machine Learning

GPA 1.1 (best: 1.0), Graduated with distinction

November 2020 – May 2023

Courses: Deep Learning, Computer Vision, Reinforcement Learning, Statistical ML

Moscow Institute of Physics and Technology

B.Sc. Applied Mathematics & Physics

GPA 4.66 (best: 5.0)

September 2016 – July 2020

Courses: Calculus, Informatics, Linear Algebra, Differential Equations, Computational Mathematics, Physics.

Projects & Research

Generative Dataset Distillation

2023 | Python, PyTorch, Wandb

Master Thesis project done under the supervision of Prof. Andreas Geiger and Kashyap Chitta
Introduced new approaches that combine dataset distillation and generative modelling
Worked with different image datasets, such as ImageNet-1k, CIFAR10/100
Used generative dataset distillation for imitation learning on CarRacing env and achieved 80+ of original score with just 1 image per class
Worked with different generative models, such as StyleGAN2 and StyleGAN-XL
Conducted extensive research of existing distillation methods and compared proposed techniques

Crypto News Telegram Bot

2022 | Python, Flask, Twitter API, Telegram API, CoinMarket API, PostgreSQL

Implemented and deployed Telegram Bot for receiving crypto news from Twitter, Discord and CoinMarket
Actively used by 250+ people and providing access to 60+ crypto-related projects.

Laser-hockey RL

2021 | Python, PyTorch, OpenAI Gym, NumPy

Implemented Deep Deterministic Policy Gradient (DDPG) and Twin Delayed DDPG (TD3) agents for laser-hockey environment
Ranked 3rd place out of 70+ participants in a tournament organized by Autonomous Learning Group @ MPI-IS

CarRacing IL & RL

2021 | Python, PyTorch, OpenAI Gym, NumPy

Implemented Deep Q-Network (DQN), Imitation Learning and Modular pipeline with geometric controller for CarRacing environment from OpenAI Gym.
Placed 3rd overall in the competition with 35+ participants organized by Autonomous Vision Group @ Uni Tuebingen .

Achievements

2024 | ECCV Workshop Best Paper Award

Our approach, GenDM, ranked 2nd place and won the Best Paper Award at the 2024 Dataset Distillation Challenge (Generative Track).

2018 | 2X Scholarship from the Foundation of Developing Innovational Education

Received a scholarship twice for being in the top 5% of students.

About Me