Ariel Kwiatkowski
Verified Expert in Engineering
Python Developer
Zürich, Switzerland
Toptal member since August 14, 2019
Ariel is an experienced developer with a strong focus on machine learning algorithms for reinforcement learning, graph analytics, image classification, and feature engineering. His physics background makes math-heavy topics a core strength, along with hands-on experience with machine learning and software engineering. Ariel has obtained his PhD in the application of AI to crowd simulation.
Portfolio
Experience
- Python - 7 years
- Machine Learning - 5 years
- LaTeX - 4 years
- Data Science - 3 years
- TensorFlow - 3 years
- Pandas - 2 years
- PyTorch - 2 years
- Flask - 1 year
Preferred Environment
GitHub, PyCharm, Windows, Linux, Automation
The most amazing...
...code I've written was the RL training setup for my thesis. It enabled rich customization for research purposes and multi-agent training with the PPO algorithm.
Work Experience
Research Engineer
Meta
- Published several papers in international journals, dealing with RL on large language models (LLMs).
- Built up the internal RL LLM codebase from scratch, facilitating multiple experiments across various parts of the team.
- Served as the technical lead of the team, making technical decisions and liaising with stakeholders from different teams.
Senior AI Researcher
AI Redefined
- Developed Cogment Lab, an easy-to-use framework for human-in-the-loop AI research and development.
- Created an AI assistant to manage a simulated high-energy grid for a big client.
- Built a system for the rail network management via voice control, using Whisper and large language models (LLMs).
Early Stage Researcher
Institut Polytechnique de Paris
- Wrote and published three peer-reviewed papers in international journals and conferences.
- Created a flexible RL library used for multi-agent crowd simulation using Python and PyTorch.
- Created a crowd simulation platform using the Unity game engine.
Research Assistant
Bitville Oy
- Built a multi-agent reinforcement learning environment in Pycolab.
- Implemented a distributed training procedure that involved training with old versions of the agent to improve ad-hoc cooperation in RLlib using TensorFlow.
- Reimplemented the above in PyTorch, plus a theory of mind learning component.
- Implemented PPO in PyTorch from scratch, with support for multi-agent environments and recurrent policies.
- Contributed to the research design by finding theoretical predictions of experiment results.
Machine Learning Engineer
Worklytics
- Developed a machine learning-based employee retention prediction algorithm based on employees' activities, including an interpretation of the model's predictions and estimating the model's certainty about its predictions.
- Performed exploratory data analysis on employee retention data.
- Explored the possibility of detecting employee positions via clustering.
Data Scientist I
CodiLime
- Implemented a feature engineering scheme to improve a network analysis-related machine learning project.
- Created a tutorial on optimization in TensorFlow.
- Coordinated with coworkers and managers to adapt results from research papers to our automated ML system.
Deep Learning Intern
SeerIT
- Implemented computer vision algorithms to detect electric insulation on images.
- Implemented a neural network for end-to-end image segmentation.
Experience
ACORES
https://github.com/RedTachyon/temperature-analysisNodeBook
https://github.com/RedTachyon/nodebook-prototypeFestival Simulation
https://github.com/RedTachyon/festival_projectACORES Label Helper
https://github.com/RedTachyon/label-helperEducation
PhD in Computer Science
École Polytechnique - Paris, France
Master's Degree in Artificial Intelligence and Robotics
Aalto University - Helsinki, Finland
Double Master's Degree (KTH, Aalto) in ICT Innovation
EIT Digital - Finland and Sweden
Master's Degree in ICT Innovation, Autonomous Systems
KTH Royal Institute of Technology - Stockholm, Sweden
Bachelor's Degree in Physics, (individual studies)
Universtiy of Warsaw - Warsaw, Poland
Certifications
Natural Language Processing Nanodegree
Udacity
Deep Reinforcement Learning Nanodegree
Udacity
Deep Learning Specialization
Coursera
Skills
Libraries/APIs
TensorFlow, OpenAI Assistants API, Keras, PyTorch, Pandas, NumPy, Scikit-learn, Node.js, OpenCV, JAX
Tools
ChatGPT, PyCharm, LaTeX, GitHub
Languages
Python, SQL, C++, JavaScript
Paradigms
Automation, Agile
Frameworks
Flask
Platforms
Google Cloud Platform (GCP), Linux, Windows
Storage
SQLite
Other
Data Science, Natural Language Processing (NLP), Deep Reinforcement Learning, Machine Learning, Artificial Intelligence (AI), Generative Pre-trained Transformers (GPT), Convolutional Neural Networks (CNNs), Computer Vision, Image Recognition, OpenAI, Large Language Models (LLMs), AI Agents, AI-generated Code, CSV File Processing, Reinforcement Learning, Robot Operating System (ROS), Research, Technical Writing
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring