ML Research Engineer @ Iliad Group | Free / Scaleway

Engineering the intelligence in Artificial Intelligence. :)

I am a Computer Scientist passionate about Artificial Intelligence, MLOps, and Data Science. After graduating from École Polytechnique at Institut Polytechnique de Paris, France, and National Polytechnic Institute (ENSEEIHT), France, I will join Iliad Group as a ML Research Engineer in Paris. My work involves developing cutting-edge ML solutions, from deep learning models in geoAI with cool stuffs to robust data pipelines. My research interests include computer vision, NLP, generative models, model compression, and efficient deep learning. I am driven by the potential of machine learning to address complex challenges and create impactful solutions.

Outside of work, I love traveling to explore new countries and cultures. I am also deeply committed to equality of opportunity. As an AI Young Leader recognized by the UN, I am leveraging AI to support individuals and communities in need, ensuring that technological advancements benefit all.

from Persistence import Excellence
from ComputerScience import ExcellentProgrammer

class BryanChen(ExcellentProgrammer):

    def __init__(self):
    
        self.country = "France 🇫🇷"
        self.current_city = "Paris, France"
        self.current_job = "AI Applied Engineer @ Iliad Group (Free / Scaleway)"
        self.past_experiences = ["ML Research Engineer @ NUS School of Computing",
                                 "ML Research Engineer @ CNRS",
                                 "ML Research Engineer @ LAAS-CNRS"]
        self.universities = ["École Polytechnique - France",
                             "ENSEEIHT - France",
                             "ENS Paris-Saclay - France"]
        self.passions = ["traveling", "exploring_cultures", "languages", "gyoza making"]
        self.volunteer_engagements = ["Young AI Leader @ ITU (Paris-Hub) for AI for Good"]

    def im_interested_in(self):
        
        return ["deep_learning", "machine_learning", "data_science",
                "computer_vision", "natural_language_processing", 
                "generative_models", "model_compression", "responsible_ai"]

    def goals_and_commitments(self):

        return self.develop_impactful_ai_solutions() \\
               and self.advance_research_in(["machine_learning", "responsible_ai"]) \\
               and self.contribute_to_open_source() \\
               and  Excellence("everything")

I'm excited to be part of the Iliad Group (Free / Scaleway) team in Paris, France, as a ML Research Engineer. My work focuses on proposing deep learning models for signal quality coverage maps, developing data pipelines and ETL, performing feature engineering, rigorous model evaluation, and building Streamlit applications querying Clickhouse databases to present key business insights.

During my time at the National University of Singapore (NUS) School of Computing (SoC) in Singapore, I worked as a ML Research Engineer. I created an efficient checkpointing fine-tuning scheme for DNNs using Delta-LoRA and LC-checkpoint, achieving compression ratios up to 25x. This was applied to models like ViTs, ResNets, LeNet-5, VGG-16, and AlexNet. (See Code)

As a ML Research Engineer at CNRS in Toulouse, France, I developed an interactive optimization algorithm for a Constraint Satisfaction Problem (CSP), applying Neural Networks and Decision Trees. This work significantly improved decision-making, achieving up to 80% of the theoretical objective function value. (See Code)

As a Young AI Leader for the International Telecommunication Union (ITU) at the Paris-Hub, I am actively involved in the United Nations' AI for Good program. Our team, the Young AI Leaders of Paris, is a group of dedicated young volunteers committed to promoting an ethical, responsible, and human-centric artificial intelligence for the common good.

We believe that while AI is rapidly transforming our societies, it remains largely misunderstood. Our mission is to democratize access to a rigorous, accessible, and deeply humanistic AI culture. We aim to bridge the gap in understanding for students, non-specialist researchers, and citizens, thereby reducing confusion and fostering engagement.

In this capacity, I collaborate with the team to:

• Engage with local organizations to understand their on-the-ground realities and needs.
• Explore how AI technology can modestly contribute to supporting their missions.
• Develop concrete projects, such as our current exploration of a solidarity chatbot for individuals experiencing homelessness or extreme precarity.

Our approach is fundamentally collaborative and field-driven, ensuring that the solutions we explore are relevant and impactful. This engagement is a core part of my commitment to ensuring technology serves humanity and contributes positively to society.

I am passionate about pushing the boundaries of AI and have actively participated in data challenges and contributed to research.

• Rank 1/178: Inria & APHP - Mean Arterial Pressure (MAP) prediction with ECG and PPG signals. (Code)
• Rank 2/148: Inria & Hi!Paris - Groundwater level prediction. (Certificate & Report)
• Rank 4/87: MVA RecVis 2024 - ImageNet-sketch classification. (Report)
• Rank 6/30: EUROSAT classification with image classification DNNs. (Notebook)

• CoVR-2 (AAAI Extension): Explored balanced, context-aware representations and better embeddings alignment. (Paper)
• Improving Vision Language Models: Extended sparse attention vectors approach. (Report)
• HYGENE: Diffusion-based Hypergraph Generation (AAAI Presentation): Hypergraph generation with diffusion. (Report)
• Impact of Knowledge Distillation for Model Interpretability (ICML Blog): Blog post on interpretability. (Blog Post)
• Classifier-Free Diffusion Guidance (NeurIPS Review): Jointly training conditional/unconditional diffusion models. (Report)
• Recommender Systems with Generative Retrieval (NeurIPS Presentation): TIGER, generative retrieval of item IDs. (Presentation)
• Predicting Naturalness using Acoustic Indices: Utilized scikit-maad, VGGish & PANNs. (Code)
• Neural Graph Generation conditioned on Text Descriptions: Project at École Polytechnique. (Report)

I've undertaken numerous projects spanning a diverse range of fields, including internships, research endeavors, academic coursework, and competitive data challenges. These are consolidated into repositories on my GitHub. You'll discover a wide variety of subjects within, including Deep Learning, Optimization, Mathematics, Software Development and Algorithms, among others!

I'm always excited to take on new challenges in AI research and application. If you have an interesting project, a research idea, or just want to discuss the latest in tech, let’s connect! I'm open to collaborations and geeking out about all things AI :)