Maxime Bonnesoeur

AI Software Enigneer - Founding team member

Jan 2025 - Present

Stealth Finance Startup, Zürich, Switzerland

• Built and automated backoffice operations for fintech industry, streamlining processes and reducing manual work
• Developed custom data extraction tools to aggregate complex financial data for regulatory compliance reporting
• Engineered text and graph extraction systems to automate document data retrieval
• Architected and deployed secure infrastructure stack for platform hosting and LLM deployment, ensuring regulatory compliance via IAC

Data Scientist

Jul 2024 - Nov 2024

Similarweb, Zürich, Switzerland

• Transition and Integration Roles:
• Actor in Transition: Transitioning predictive systems from 42Matters to be integrated into Similarweb's future products, ensuring seamless integration and continuity.
• System Maintenance and Empowerment: Post-acquisition, responsible for maintaining and teaching about existing data science systems to the wider teams at Similarweb, while re-implementing and enhancing their current systems to leverage advanced predictive capabilities.

Data Scientist

Jul 2022 - Nov 2024

42Matters - A similarweb company, Zürich, Switzerland

• 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 𝗧𝗲𝗮𝗺: Led and developed the data science team, empowering and mentoring members and growing with them. Spearheaded project formulation and execution, ensuring successful machine learning model deployment and timely delivery.
Notable projects and contribution
• App Categorization: Developed an advanced IAB classification system leveraging text embedding, cosine similarity, and Amazon Mechanical Turk. This initiative doubled the coverage of the previous system and increased accuracy by 25%, significantly improving ad targeting and blacklist detection.
• Revenue Estimation: Created sophisticated models to estimate app earnings, enabling customers to pinpoint high-engagement apps. This provided actionable insights for better-targeted marketing strategies and informed business decisions.
• App Matching: Engineered a graph-based system using Spark and computer vision for efficient cross-marketplace app matching. This solution facilitated accurate comparisons and integrations across different app marketplaces.
• Demographics Estimation: Implemented demographic estimation through computer vision and Spark, offering clients detailed user profile insights. This enhanced the ability to conduct targeted marketing campaigns based on age and demographic data.
• Integrated databases and Augmented Generation RAG search for easy review analysis and app discovery

Junior Machine Learning Engineer

2021 - 2022

Visium, Lausanne, Switzerland

Visium is an AI and data-based consulting company which integrated the top 2021 10 best startups in Switzerland - Linkedin

• Created a NLP model to identify the preferred incentives of the Swiss political class toward some predetermined domains.
• Creation of a web application for the study and automatic detection of false watches based on the analysis of their vibration's signature.
• Development of orchestration pipelines via Airflow to setup Virtual machines and network environment (Terraform) / Slack channels / Github projects / Hour tracking of projects to fully monitor the resources and expenses involved in their development.
• Designed and Implemented the sale and event tracking architecture of a legal company using AWS, Snowflake and DBT. The implementation aggregates the events generated by the platforms into sessions and users which lowered the costs of processing ( by a factor of 5) and produced significant insights for the analyst's team.

Master of Robotics, specialised in deep learning

2018 - 2021

EPFL, Lausanne, Switzerland

Master oriented toward machine-learning techniques applied to autonomous systems and to develop continuous improvement workflows.

Thesis name: Monocular 3D vehicle detection and distance estimation with self-attention mechanism

Bachelor of Micro-engineering

2015 - 2018

EPFL, Lausanne, Switzerland

Learned programming and the basics of micro-engineering and mechanical engineering to build complex products.

VITA laboratory

2020-2021

VITA laboratory, Lausanne, Switzerland

Visual Intelligence for Transportation (VITA) is a laboratory publishing at renowned competitions (CVPR, NeurIPS) to advance the state-of-the-art in the domain of autonomous transportation.

• Developed an attention-based mechanism applied for human and vehicle detection and 3D localization from a monocular RGB image (no lidar or stereo images involved). This technique originally developped for Transformers in NLP was here implemented for visual tasks.

Intern Digital Engineering for Industry 4.0

2020

Philip Morris International, Neuchâtel, Switzerland

• Proposed data processing techniques to the management team to detect defects and adapt the machine's speed to produce the HEETS (tobacco sticks for the IQOS).
• Worked on a new parameter adaptation solution that reduces the number of stops by 80% and produces a product of better overall quality for the end-user.

Machine Learning Intern

2019

Rovenso, Villaz-Saint-Pierre, Switzerland

• Implemented a machine learning model to detect humans with both thermal (FLIR) and RGB cameras .
• Optimised the network to work on an NVIDIA Jetson. The backbone of this project is still nowadays used in the main product of the company.