Summary of data and AI projects we actually made
It may look like a resume. In fact, it is. And we promise you it's not just talk. Either we did these projects ourselves with our bare hands, or we led the team of data scientists who carried them out. For real.
let's summarize it
We specialize in leading data strategy and building AI-powered tools that make a real impact across various industries. Our work includes creating smart anomaly detection systems, optimizing machine learning models, and developing algorithms that classify and analyze data more effectively. We’re passionate about using data to help businesses make better decisions, streamline their operations, and unlock new opportunities. With a focus on healthcare, fintech, and industrial sectors, we use advanced analytics and AI technologies to solve complex problems and drive innovation.
Booooooring
Actually we do like data science.
As far as we are concerned, data science and AI have nothing to do with LLM prompting.
2024 PROJECTS
Designed and implemented anomaly detection programs for a startup, focusing on waste detection in urban and natural environments. Developed deep learning models tailored for small datasets from drone and fixed camera images. Enhanced environmental monitoring through AI-driven insights.
Leading data strategy and technology for a startup, building a scored database of scientific and medical knowledge in immunotherapy. A founding member of the Marseille Immunology Biocluster (MIB)), which secured 97M€ in funding from France 2030.
2023 PROJECTS
Developed a statistical learning tool for clinical study data analysis in a medical biotech company. Improved data processing efficiency and provided insights to enhance research and decision-making.
Identified key parameters influencing defects in a production line for an industrial company. Used data science techniques to optimize quality control and reduce manufacturing waste.
Supported the internal data team of a biotech research company. Co-designed digital tools to enhance data analysis and improve research protocol efficiency.
Conducted a technical audit of machine learning algorithms for damage detection for a company specializing in automotive technology. Assessed model performance and provided optimization recommendations.
Performed a statistical study on the correlation between biomarkers and hearing function in an animal model for a biotech research company. Delivered key findings to guide future studies.
2022 PROJECTS
ChatGPT had just been released and LLMs were all the rage at the end of 2022. Prepared a technical note on LLM technologies for C-level executives at a global shipping company. Provided insights on AI adoption and strategic implications. They decided not to pay us, now I hate them.
Developed an automatic detection method for children falling in swimming pools for an SME in safety technology. Designed and prototyped a decision-making tool to enhance accident prevention. It worked OK in the lab.
Trained a team of 15 statisticians in applied mathematics and machine learning for a global healthcare data company. Strengthened in-house expertise through hands-on learning and real-world applications.
Conducted a research study for an SME in digital marketing, evaluating potential external acquisitions in data science. Provided strategic insights for investment decisions.
Developed an algorithm for automatic classification of interest declarations (short texts) for a government transparency agency (HATVP). Designed, prototyped, implemented, and deployed the system into production. We now consider including LLM as another metric. Our first project powered by Streamlit.
Designed and deployed an anomaly detection tool for a partner specialised in database meta-engine. Predicted storage and CPU usage on servers, improving IT infrastructure monitoring and decision-making.
Conducted a feasibility study for a SaaS company on optimizing eco-logistics routes for last-mile delivery. Explored operational research approaches for improved efficiency. The exec team discovered the concept of NP-hardness through this study and was very angry about that.
Aggregated and cleaned databases for an agricultural cooperative and investment fund. Improved data quality and usability for business intelligence applications.
Performed data mining on 10 million medical studies for a SaaS in media. Extracted insights to enhance research accessibility and knowledge discovery. So much databases.
2021 PROJECTS
Updated and retrained an automatic classification engine for a government transparency agency HATVP, improving accuracy and efficiency in processing declarations of interest.
Developed a banking transaction classification system for a fintech company, enhancing automation and compliance in financial operations. We like them.
Conducted an internal risk assessment for a global shipping company, contributing to the 2022 audit plan with data-driven insights. Like running the 2021 program again with updated data.
Performed a technical audit of a machine learning engine for a SaaS company, assessing model performance and reliability. Did they solve the right problem?
Calculated recommendation engine performance metrics for a SaaS company, optimizing personalization algorithms. Still wondering why their top-notch team was not able to do that.
Built a knowledge base and recommendation engine for the French education system, guiding training and career path selection for a SaaS company.
Implemented non-compliance detection algorithms in internal audit procedures for a global shipping company, strengthening governance and regulatory oversight. I could speak a lot more about this project, unfortunately I had to sign a NDA with my own blood.
Designed and implemented data processing routines (statistics, ETL) for a fintech company, streamlining data workflows and analytics capabilities.
2020 PROJECTS
Conducted an internal risk assessment for a global shipping company, contributing to the 2021 and 2020 audit plan with data-driven insights for improved governance. We processed a lot of xls files, yes we did..
Designed a scientific article recommendation engine for a SaaS company, creating a structured knowledge base of articles, authors, and research laboratories.
Developed an automatic expert system for a fintech company, leveraging the SIRENE public database to enhance decision-making and business intelligence.
2019 PROJECTS
Developed a price calculation tool for an e-commerce company, integrating a sales model tailored for 15 online platforms to optimize pricing strategy. This tool was a perfect fit for streamlit, still not in our radar at that time.
Performed data mining and automatic classification for a nuclear safety institute, using wavelet-based time-frequency analysis on reduced datasets. Was fun to show how ML works to nuclear engineers.
Identified mechanical faults in helicopter test bench data, detecting unknown frequency signatures in raw sensor data. I could speak a lot more about this project again, unfortunately I had to sign a NDA with my child's blood.
Designed an automated tax return processing system for a regional bank, implementing client-product matching and recommendation rules. That was actually very easy, hope they will not read this post.
Led a scoping mission with the senior VP of audit & risk at a global shipping company, defining key risk management priorities. Our first mission within the highest skyscraper of Marseille.
Developed an automatic classification tool for lobbyist declarations at a government transparency agency, using statistical modeling techniques.
Designed, prototyped, and deployed a machine learning-based classification engine for banking transactions in a fintech company, improving transaction categorization accuracy.
2018 PROJECTS
Conducted a technical audit of the data and ML team for a fintech company, assessing workflows, model performance, and best practices. "What the hell is our ML team doing?" was the core question of this mission. Now they know.
Provided operational data coaching for RATP executives, enhancing their understanding and application of data-driven decision-making. I led many interviews at StationF in Paris, so I liked this mission.
Performed MBA courses on data science and AI at ESG Paris, covering three different levels of expertise. They knew nothing about computer science.
Led a commercial and technical strategy mission for a large opinion pool company, focusing on the monetization and optimization of internal data assets. They were just lost at that time.
Facilitated ideation workshops for BNF Partenariat, exploring strategies to enhance the Retronews documentary collection through data-driven insights.
Offered data and digital support to a SME franchisor, optimizing business operations through technology and analytics.
Performed a strategic audit of the data and AI roadmap for a startup, refining the company's approach to artificial intelligence and data-driven growth. I told them they could not so heavily rely on human operation if they really wanted to scale - the CEO was not convinced. Hope he changed his mind.
Presented a conference on blockchain technology for insurance company, engaging an audience of 100 professionals in Paris. Much fun.
2017 PROJECTS
Provided operational data coaching for bank executives, helping them integrate data-driven strategies into decision-making.
Developed and deployed a web scraping and keyword extraction system for an SME, processing 1.8 million websites to extract relevant insights automatically. Might have been illegal.
Offered technical and methodological support to a fintech data analyst, improving analytical processes and data handling.
Authored a research project for an innovative logistics startup, including algorithmic design for advanced operational efficiency. Had to explain NP-hardness another time.
Delivered data training for Stellantis Group executives, enhancing their understanding of data-driven business strategies.
Designed a decision-making tool for the French Environment and Energy Management Agency (ADEME), classifying and visualizing building typologies across France at scale. Much fun, a lot of data and the project was really nice with high dimensional data.
2016 PROJECTS
Advised executive managers of a retail firm on data strategy, helping define key initiatives for leveraging data in business operations.
Led the recruitment of a data analyst for a fintech company, assessing candidates' skills and expertise.
Built and supported a data science team for a fintech company, structuring workflows and methodologies for analytics projects.
Performed data mining on unstructured data from the SIRENE database for an SME, extracting valuable business insights.
Developed outlier detection models for the French Environment and Energy Management Agency (ADEME), analyzing millions of French buildings.
Designed an automatic classification system for accounting transactions using supervised machine learning for fintech company, delivering a proof of concept.
Spoke at Hello!World 2016 for Orange Business Services in Lisbon, presenting to an audience of 300 professionals. Much stress.
Developed statistical analysis plans for a biotech company, analyzing pharmaceutical and veterinary data for research and market insights.
Built keyword extraction and unstructured data processing tools for an SME, processing 100,000 websites from design to production.
2015 PROJECTS
Our first year as consultant! We started working for companies as data scientists when people got excited about Big Data and noSQL.
Analyzed customer attrition for a fintech, identifying behavioral patterns through statistical and semantic analysis of customer feedback to improve retention strategies.
Conducted a technical assessment of data scientist candidates for a startup in Paris, evaluating their skills and expertise in machine learning and data analysis.
Designed and developed a commercial segmentation tool for fintech, leveraging banking data from 300,000 customers to enhance marketing strategies.
Automated Excel file processing for an SME industry, implementing VBA and Python scripts to streamline data handling and improve efficiency. Will not do it anymore, I have VBA so much.
Did you really do all of this?
Even more actually.
Thanks to ChatGPT and DeepL, I would never have had the courage to write this post without you, even in French