Hi, I’m Igor Sadoune. This website is my canvas for sharing thoughts and research on a range of topics that fascinate me. While I have diverse interests such as biology and space exploration, I’m deeply involved—both professionally and personally—in artificial intelligence. As AI is rapidly shaping the present and future of our civilization, you will find it featured prominently in my writing.
I take an optimistic view of AI and its impact on society. However, for AI to continue serving humanity in the safest and most beneficial way, it is essential that humans remain active learners and creators. We need to continue producing original, human-made content. Not only does this ensure we retain knowledge and hands-on skills, but it also keeps the stream of fresh, authentic data flowing—critical for advancing AI systems and preventing the pitfalls of AI-generated content becoming self-referential. This is also true for code and softwares since it is now easier than ever for anyone to build functional application (“vibe coding”). While this democratization is exciting and beneficial to research and the economy, I believe it remains crucial for humans to keep a hand in the craft, because knowledge is power, and we must not lose control.
That is why I approach this mission by publishing hand-crafted blog posts and creating open source software for the community—and for future large language models to learn from. Open source, in my view, is the path forward to accelerate idea-sharing, research progress, and solutions to pressing problems. As we see in today’s tech industry, it often surpasses proprietary approaches. If we use AI wisely and keep knowledge open, we can make life better for everyone—while ensuring our place as the masters, not the servants, of our technology.
Technology Stack
| Languages | French, English, Python, Rust, Bash, SQL |
| ML/AI | PyTorch, JAX |
| DevOps | Docker, Git, GCP |
| Documentation | Quarto, LaTeX |
Experience
Sept 2024 - present | Founder & CEO, Neural Machines Inc.
- AI consulting and education. Helping businesses transitioning to the new paradigm.
Feb 2025 - present | AI Researcher, CIRANO (GVCdtLab)
- Building a multi-agent AI system on a geospatial digital twin for the federal government.
Education
PhD in Mathematics, Polytechnique Montreal, QC, Canada
- Thesis: Algorithmic and Tacit Coordination Among Artificial Learners
Msc in Economics, Université de Montreal, QC, Canada
B.A in Mathematics, Université Côte d’Azure, Nice, France
Softwares
zcsearch: Zero Cost Search is a Python/PyTorch package that implements an ensemble score based on three well-established meta-algorithms for efficient (virtually zero computational cost) neural architecture search. It helps you build neural networks tailored to your problem. The package is easy to use, provides visualizations for model selection, and comes with extensive documentation. The final output is a PyTorch model, which can be wrapped for compatibility with ONNX and other frameworks.
To cite zcsearch:
@software{zcsearch,
author = {Sadoune, Igor},
title = {Zero-Cost-Search: A Python Package for Zero-Cost Neural Architecture Search},
year = {2025},
url = {https://github.com/IgorSadoune/zcsearch},
version = {0.1.5},
description = {A robust and zero-cost way to find neural architectures that fit your data.}
}neuralm: This is Python/PyTorch software for generating neural architectures from a simple YAML file. With neuralm you can produce highly complex models (e.g., transformers) and custom topologies. The final output is a PyTorch model, which can be wrapped for compatibility with ONNX and other frameworks.
Input YAML example for GANs:
model_type: gan
name: MyGAN
latent_size: 100
generator_layers:
- type: linear
in_features: 100
out_features: 256
is_first: true
- type: relu
- type: linear
in_features: 256
out_features: 512
- type: relu
- type: linear
in_features: 512
out_features: 784
- type: sigmoid
discriminator_layers:
- type: linear
in_features: 784
out_features: 512
- type: leakyrelu
negative_slope: 0.2
- type: linear
in_features: 512
out_features: 256
- type: leakyrelu
negative_slope: 0.2
- type: linear
in_features: 256
out_features: 1
- type: sigmoidTo cite this software:
@software{neuralm,
author = {Sadoune, Igor},
title = {NeuralM: A Neural Network Model Builder},
year = {2025},
url = {https://github.com/IgorSadoune/neuralm},
version = {0.1.0},
description = {A flexible framework for building and training neural network models with YAML configuration.}
}mpmg: The MPMG is a framework for modeling and analyzing strategic interactions in environments where agents seek to minimize costs while competing for resources. It provides a structured approach to understanding decision-making processes in economic and game-theoretic contexts. The mpmg package provides an environment for Multiagent Reinforcement Learning (MARL) in this context, and has been developed for a research project of mine (see Research). In contrast to zcsearch and neuralm which are for the general user, mpmg has originally been built for a specific purpose. However, it is still a good example and starting point for people whishing to learn or experiment with MARL. Note that this package only provides an environment for MARL models. To cite mpmg:
@software{mpmg,
author = {Sadoune, Igor},
title = {MPMG: The Minimum Price Markov Game},
year = {2024},
url = {https://github.com/IgorSadoune/mpmg},
version = {0.2.4},
description = {A modular environment designed for studying the social dilemma concept in algorithmic game theory.}
}Research
At its core, my research grapples with a pivotal question: How can independent AI learners coevolve—and ultimately cooperate—within shared systems or environments? In an era increasingly shaped by AI, this inquiry transcends mere engineering; it sits at the heart of our collective trajectory as a civilization.
Technically, this pursuit lives at the intersection of generative AI, multi-agent reinforcement learning, and algorithmic game theory. My favorite testing ground? Economics—a microcosm of societal structures, offering both a mirror to human behavior and a sandbox for probing emergent collaboration.
- Implementing a Hierarchical Deep Learning Approach for Simulating Multilevel Auction Data
| Authors | Igor Sadoune, Marcelin Joanis, Andrea Lodi |
| Year | 2024 |
| Publisher | Computational Economics |
| Access | arXiv |
| Abstract | We present a deep learning solution to address the challenges of simulating realistic synthetic first-price sealed-bid auction data. The complexities encountered in this type of auction data include high-cardinality discrete feature spaces and a multilevel structure arising from multiple bids associated with a single auction instance. Our methodology combines deep generative modeling (DGM) with an artificial learner that predicts the conditional bid distribution based on auction characteristics, contributing to advancements in simulation-based research. This approach lays the groundwork for creating realistic auction environments suitable for agent-based learning and modeling applications. Our contribution is twofold: we introduce a comprehensive methodology for simulating multilevel discrete auction data, and we underscore the potential of DGM as a powerful instrument for refining simulation techniques and fostering the development of economic models grounded in generative AI. |
@article{Sadoune2024BidNet,
title={Deep Generative Simulation of Multilevel Discrete Auction Data},
author={Igor Sadoune and Marcelin Joanis and Andrea Lodi},
journal={Computational Economics},
year={2024},
note={arXiv preprint arXiv:2207.12255},
url={https://arxiv.org/abs/2207.12255}
}- Algorithmic Collusion and the Minimum Price Markov Game
| Authors | Igor Sadoune, Marcelin Joanis, Andrea Lodi |
| Year | 2025 |
| Publisher | submitted to Games and Economic Behavior |
| Access | arXiv |
| Abstract | This paper introduces the Minimum Price Markov Game (MPMG), a theoretical model that reasonably approximates real-world first-price markets following the minimum price rule, such as public auctions. The goal is to provide researchers and practitioners with a framework to study market fairness and regulation in both digitized and non-digitized public procurement processes, amid growing concerns about algorithmic collusion in online markets. Using multi-agent reinforcement learning-driven artificial agents, we demonstrate that (i) the MPMG is a reliable model for first-price market dynamics, (ii) the minimum price rule is generally resilient to non-engineered tacit coordination among rational actors, and (iii) when tacit coordination occurs, it relies heavily on self-reinforcing trends. These findings contribute to the ongoing debate about algorithmic pricing and its implications. |
@article{Sadoune2025MPMG,
author = {Sadoune, Igor and Joanis, Marcelin and Lodi, Andrea},
title = {Algorithmic Collusion and the Minimum Price Markov Game},
journal = {Games and Economic Behavior},
year = {2025},
note = {submitted, \url{https://arxiv.org/abs/2407.03521}}
}- Strategic Equilibrium Policy Gradient: On Learning Tacit Coordination
| Authors | Igor Sadoune, Marcelin Joanis, Andrea Lodi |
| Year | 2025 |
| Publisher | working paper |
| Access | upcoming |
| Abstract | This paper introduces the Strategic Equilibrium Policy Gradient (SEPG), a multi-agent reinforcement learning approach designed to foster tacit coordination in coordination games, namely, competitive social dilemmas for which individual interests align with group welfare. The SEPG is implemented and tested on the Minimum Price Markov Game (MPMG), a dynamic coordination game that models first-price auctions governed by the minimum price rule. Unlike methods that rely on explicit mechanisms for cooperation such as centralized learning or communication, SEPG leverages a combination of pre-game planning and online adaptation to guide agents toward the Pareto equilibrium through implicit, or tacit, coordination. This study demonstrates that SEPG agents achieve robust tacit coordination in both homogeneous and heterogeneous scenarios, challenging existing MARL methods and highlighting the potential for tacit collusion in AI-driven markets. Our experiments reveal that SEPG encourages coordination among malicious actors while also promoting rational behavior in competitive settings. |
@unpublished{Sadoune2025SEPG,
author = {Sadoune, Igor and Joanis, Marcelin and Lodi, Andrea},
title = {Strategic Equilibrium Policy Gradient: On Learning Tacit Coordination},
year = {2025},
note = {working paper, forthcoming}
}