Baptiste Marc Mokas

As a mentor for data science students, I am dedicated to inspiring a passion for data science and highlighting its transformative potential. I provide essential guidance on complex concepts, algorithms, and statistical techniques to foster academic growth. By collaborating on projects, I guide students through design, data collection, analysis, and presentation, offering hands-on experience. I bridge academia with real-world insights on industry trends, best practices, and practical applications to prepare students for the workforce. Engaging in reciprocal learning, I evolve both personally and professionally as a mentor and data scientist. I am excited to continue impacting the growth and development of aspiring data scientists.

Part Time professor of mathematics including epistemology, systems theory, stochastic dynamics and probability, descriptive statistics, exploratory data analysis, inference and estimation theory, linear statistical models, non linear models and non parametric statistics

As a participant in the Acceleration Apprenticeship Program at Europe's largest startup incubator, I had the opportunity to immerse myself in a dynamic and innovative environment. This experience allowed me to work closely with emerging startups, contributing to their growth and development through strategic mentorship and hands-on support. I honed my skills in project management, business development, and data-driven decision-making, while also gaining valuable insights into the latest technological advancements and industry trends. This program equipped me with a robust understanding of the startup ecosystem and enhanced my ability to drive innovation and success in fast-paced, high-stakes environments.

Video production on youtube

DECATHLON | TECHNOLOGY – SUPPLY CHAIN FORECASTING | 2024 Senior Lead Data Scientist – Forecasting, Supply Chain & R&D 12 months – Paris, France Development of predictive models for demand forecasting, stock optimization, and replenishment. Automation of end-to-end pipelines, model monitoring, and CI/CD. Context: large-scale retail, omnichannel e-commerce, multi-warehouse supply chain, strong seasonality, logistical constraints (lead time, capacity, cost). Key goals: multi-scale forecasting (SKU to warehouse, daily to seasonal), probabilistic predictions (quantiles, confidence intervals), incremental learning with time cutoffs, reduction of stockouts and overstocks via dynamic adjustments. Stack: Python (PySpark, Scikit-learn, TensorFlow), FastAPI, Streamlit, Poetry, Databricks, AWS S3, Airflow, Delta Lake, Parquet, MLflow, Docker, GitHub Actions, Tableau.

As a Data Scientist, I designed machine learning models to categorize scientific articles and assign a probability score for inclusion in systematic literature reviews (SLR). The goal was to develop an application employing web scraping to retrieve data from PubMed and various scientific journals, processing and prioritizing them efficiently. This project, conducted in a medical context, aimed to streamline systematic literature reviews for medical device evaluations, ultimately saving time for the responsible individuals.

As a Student Intern in Entrepreneurship for four months, I immersed myself in the foundational aspects of creating and managing data science-driven businesses. My experience encompassed developing and analyzing business models, managing finance, and ensuring accountability. I conducted extensive research on user experience (UX) design to create intuitive and effective interfaces for data science applications. This role also involved applying principles of entrepreneurship to identify market opportunities, designing scalable solutions, and validating business ideas through rigorous testing and feedback. My hands-on training in these areas equipped me with a comprehensive understanding of how to build and sustain successful ventures in the data science industry.

In the realm of cybersecurity, I undertook a mission involving the comparison between traditional directed Bayesian models and undirected probabilistic graphical models, such as Markov models. By evaluating and contrasting both approaches, I strategically pivoted towards undirected models. This shift allowed for a nuanced exploration of cyber threats, leveraging the flexibility and adaptability inherent in undirected models like Markov networks. How is information integrated ? How to determine the phase-space and the dynamic of multi-objective optimization systems ? How to understandlocal functions sets interactions; how to integrate their locals dynamics to understand their globals dynamics ? We propose a general geometric framework, to explain the integration of information present behind the adaptive optimization process in distributed complex adaptive organic or artificial multi-objective systems in the context of heterogeneous and dynamical environment, for modeling and simulate complex and non-linear dynamics that occurs in such multi-scale structures. We propose the building of a theoretical framework to mathematize the evolutive algorithmic multi-objective optimization. In this perspective, we seek to understand how a co-evolving and codetermined system with an heterogeneous and dynamical environment, composed of several instances, applies to synchronize its different component -its different functions sets - in a coerant, useful, synergistic, syntonic network of interaction, allowing the optimization of the fitness. We propose to geometrize the information endogeneisation - the perception / action cycle - and make a mathematical sens of reinforcement feed-backed process, with the perspective to understand the emergence of cognitive complexity, its recursiveness, autopoietic and meta-dynamical characteristics.

During my Advanced Data Science Scholarship at the New York Data Science Academy (NYCDSA), I gained in-depth expertise in a wide range of data science skills. The program emphasized practical, hands-on experience with big data management, advanced data visualization techniques, and comprehensive knowledge of data structures. I became proficient in R and Python programming, enhancing my capabilities in machine learning and data analytics. Additionally, I developed strong scripting skills and advanced my understanding of SQL for database management. This intensive training has equipped me with the tools and knowledge necessary to tackle complex data challenges and drive impactful insights in various data-driven projects.

- Theories and modeling of complex systems - Social simulations - Multi-agent programming with NetLogo - Network theory and analysis - Chaos theory and nonlinear dynamics - Emergence and self-organization - Fractal geometry - Cellular automata - Game theory - Computational complexity - Information theory - Adaptive systems - Cybernetics - Agent-based modeling - Complexity metrics and measures - Dynamical systems theory - Self-similarity and scale invariance - Evolutionary algorithms and genetic programming - Robustness and resilience in complex systems - Interdisciplinary collaboration and integration

In my machine learning project within the social sciences, I focused on modeling physiological and physical behaviors using motion capture technology and Riemannian projection techniques. The objective was to analyze and interpret human motion data to understand behavioral patterns and physical responses in various social contexts. By leveraging advanced machine learning algorithms and motion capture data, I aimed to create accurate models that represent the intricate dynamics of human movement and physiological states. This project provided valuable insights into human behavior, contributing to the broader understanding of social interactions and physical responses.