CV – Nolan Cacheux

Professional Experience

Decathlon FranceSept 2025 - Present

AI & ML Engineer (Work-Study)

Developed DAISI (Decathlon AI Suppliers Information): an AI chatbot deployed on Google Chat, providing instant answers about supplier processes, trained on internal procedures, with escalation capabilities to accountants. Estimated impact: 13,000 hours/year saved for the suppliers team.
Developing production-ready RAG architectures using LangChain, LangGraph, and Vertex AI Vector Search with GCP Cloud Run, Terraform IaC, and 24/7 availability with automated GDPR-compliant data cleanup and retention policies.
Implemented MLflow experiment tracking, LLM-as-Judge evaluation pipeline, Model Armor anti-prompt injection, and OpenTelemetry distributed tracing for full observability.

Technologies: LangChain, LangGraph, RAG, Gemini, Vertex AI, GCP, Cloud Run, Terraform, FastAPI, MLflow, Docker, PostgreSQL

Decathlon BelgiumMay 2025 - Aug 2025

Data Scientist - MLOps (Internship)

Led a strategic Data Science project to industrialize the prediction of 8 key sales KPIs (GMV, items sold) segmented by channel (InStore/OutStore, 1P/3P) for 64 sports categories.
Built an automated forecasting pipeline using Prophet, Apache Spark, and Databricks. Achieved +15% forecast accuracy improvement (MAPE) vs. previous manual process through rigorous model comparison (Prophet, XGBoost, LightGBM, Chronos-Bolt) and feature engineering (weather, holidays, lag features).
Implemented parallel processing with joblib and deployed models to MLflow Model Registry for production inference. Automated exports to Google Sheets for business stakeholders.

Technologies: Prophet, Apache Spark, Databricks, MLflow, AWS S3, Airflow, GitHub Actions, Python

Beobank NV/SAMay 2023 - Aug 2023

Data Analyst/Scientist - Banking Domain (Internship)

Mastered Python, VerticaPy, SQL, and data processing technologies. Developed analytical dashboards and reports using Matplotlib for business intelligence and decision support.

Technologies: Python, SQL, Pandas, NumPy

Personal Projects

Mistral AI Hackathon London 2026 — Ecotopia GitHub

Selected among 7,000+ applicants. Organized by Mistral AI, Iterate, sponsored by W&B, NVIDIA, AWS, ElevenLabs, Hugging Face. Built an interactive political simulation powered by fine-tuned SLMs. 4 models via QLoRA (NF4 4-bit, r=16), <10 min/model. 8B outperforms Mistral Large on structured output at 10x lower latency.

Stack: Mistral, QLoRA, BitsAndBytes, HuggingFace, W&B, Spring Boot, Spring AI, Phaser 3, TypeScript, PostgreSQL

RAG Equity Research Agent GitHub

Production-deployed multi-agent system for automated equity research. Scalable hybrid RAG pipeline on SEC filings with semantic reranking and Telegram bot interface. Deployed on Azure with Terraform IaC.

Stack: LangGraph, LangChain, Qdrant, Groq, Azure OpenAI, FastAPI, Pydantic, Terraform, Docker, Pytest

AI Video Comment Analyzer GitHub

Full-stack ML application: automated BERT sentiment classification, BERTopic topic modeling, and LLM-powered summaries. Real-time SSE streaming and interactive analytics dashboard.

Stack: BERT, BERTopic, Transformers, Ollama, Next.js 15, React 19, FastAPI, SQLAlchemy, Tailwind CSS, Recharts, Pytest

AI Product Photo Detector GitHub

Production-ready MLOps pipeline: transfer learning with EfficientNet-B0, automated tracking with MLflow, 6-step Vertex AI Pipeline, real-time drift monitoring with Prometheus/Grafana, and CI/CD with GitHub Actions.

Stack: PyTorch, EfficientNet-B0, FastAPI, MLflow, Vertex AI Pipelines (KFP), DVC, Prometheus, Grafana, Docker Compose, GitHub Actions

Distributed Data Platform GitHub

Scalable data infrastructure with Kubernetes + Helm orchestration. Enterprise-grade polyglot persistence: Neo4j, Cassandra, Redis, MinIO. Built with NestJS and TypeORM.

Stack: Kubernetes, Helm, NestJS, TypeORM, PostgreSQL, Neo4j, Cassandra, Redis, MinIO, Docker, TypeScript

Education

JUNIA ISEN - Lille, France2021 - 2026

Master's in Engineering (Diplôme d'Ingénieur), Computer Science (2023 - 2026) · Specialization in Data Science & Machine Learning.

Study Topics: Data Structures & Algorithms, Machine Learning, Deep Learning, Distributed Systems, Database Management (SQL & NoSQL), Operations Research, Computer Networks, Big Data (Hadoop, Spark), Cloud Computing, DevOps, Computer Vision, NLP, Software Architecture

Preparatory Classes · Computer Science & Networks (2021 - 2023) · Ranked 2nd/96 (Year 1) • 8th/76 (Year 2) · Highest Honors

Technical Skills

LLMs & GenAI: LangChain, LangGraph, Groq, Llama, Gemini, OpenAI / GPT, Ollama, Transformers (HuggingFace), RAG

Machine Learning & Data Science: Python, PyTorch, TensorFlow, scikit-learn, XGBoost, LightGBM, Prophet, BERTopic, Pandas, NumPy

MLOps & Orchestration: MLflow, Vertex AI, Apache Airflow, OpenTelemetry

Vector Databases: Qdrant, Vertex AI Vector Search, FAISS, Chroma DB, Pinecone

Cloud Platforms: GCP, Cloud Run, Cloud SQL, Cloud Storage, Cloud Scheduler, Vertex AI, AWS, S3, SageMaker, Bedrock, Azure, Container Apps

Big Data & Data Engineering: Apache Spark / PySpark, Databricks, Delta Lake, SQL, PostgreSQL, MongoDB, Redis, Neo4j, Cassandra

Backend & APIs: FastAPI, NestJS, Express.js, Socket.IO, Pydantic, TypeORM

Infrastructure & DevOps: Terraform, Docker, GitHub Actions (CI/CD), SonarCloud, Kubernetes, Helm

Frontend & Full-Stack: React.js, Next.js, TypeScript, Tailwind CSS, Three.js

Methodologies: Agile/Scrum, Jira, Git Flow, Code Review, Technical Documentation

Languages

French: Native|English: Professional (B2 TOEIC)

Extracurricular - Freestyle Football Champion

Professional athlete, 30K+ followers. President of Nolan-Free Association since 2019 : performances, workshops & media appearances (M6, France 3).

Junior World Champion (Prague, 2019) · 2x Vice-Champion of France (2024, 2025)
Top 6 Europe (2024) · Top 15 World (2026) · 300+ performances : LOSC Lille, Valenciennes FC, Ferrero

nolanfree.fr