Yinghua Lernould

- Experienced AI Engineer, Data Scientist, and Data Engineer with a strong focus on designing and deploying responsible, production-grade Generative AI solutions. Expertise spans advanced LLM and multimodal systems, RAG pipeline architecture, vector databases, and frameworks such as LangChain for scalable, intelligent applications for CAC40 clients. - Combines deep technical knowledge with a commitment to ethical AI, aligning innovation with business strategy, regulatory standards, and real-world impact. Dedicated to transforming cutting-edge AI research into reliable, explainable, and high-value systems across diverse industries. • AI Engineer | Enterprise LLM Application | TotalEnergies Digital Factory (TDF) - Mission: designed and delivered a production-grade application, including multilingual chatbot, intelligent document retrieval, and AI-assisted risk identification to improve on-site safety operations. - Ownership: led LLM application development from PoC to production, including system architecture, model selection, prompt engineering, evaluation, monitoring, and real-time streaming. - Impact: achieved 84% user satisfaction across 10+ countries; Real life impact on life saving; defined the client’s Gen AI technical and evaluation standards; delivered 20+ executive-level AI enablement sessions. • Tech Lead | Azure Databricks Migration | ACC - Mission: Migrated legacy on-premises ETL pipelines to a Medallion Architecture on Azure Databricks. - Ownership: Resolved complex PySpark performance bottlenecks, implemented best practices, and provided technical guidance to junior developers. - Impact: Reduced data processing latency by 40% through PySpark optimization, improved ETL efficiency, and accelerated team delivery velocity by streamlining workflows and mentoring developers. • Data Scientist | AML Transaction Monitoring | CACIB - Mission: Deployed and maintained a global AML detection system across 10+ countries, leveraging SWIFT, SEPA, and instant payment data to detect anomalous transactions and support compliance operations. - Ownership: Managed end-to-end deployment of unsupervised models across Dev, Staging, Preproduction, and Production; collaborated with business stakeholders to optimize performance and resolved production emergencies. - Impact: minimized false-positive alerts.

Delivered cross-domain AI and cloud data solutions for global clients, spanning RAG system architecture, Generative AI prototyping, and large-scale Azure data platform migrations. Led technical design and stakeholder scoping. • AI Architect | AI-powered Document Consistency Checker | International Criminal Court (ICC) - Mission: designed the AI architectural extensions for a legal intelligence platform, enabling LLM-powered inconsistency detection within a single document or across heterogeneous documents and vehicle image classification. - Ownership: scoping with stakeholders; led technology selection and the strategy for inference cost optimization. - Impact: delivered a validated the integration layer between RAG pipelines and the legacy legal system within a 2-month PoC; proved the feasibility of semantic inconsistency detection in complex legal contexts. • AI Engineer | Generative AI R&D Initiatives | Avanade Tech Office - Mission: pioneered early exploration of LLM and RAG to support client pre-sales. - Ownership: researched and prototyped RAG systems; contributed reusable code to shared repositories and collaborated with the Tech Office. - Impact: delivered the company’s first LLM-powered client demos, including email and client communication, content categorization, chatbots, and image generation. • Data Engineer | Azure Cloud Migration | AXA - Mission: large-scale migration of multiple on-premises data platforms to Azure Databricks, enabling a modern lakehouse architecture with Bronze, Silver, and Gold layers for analytics and streaming use cases. - Ownership: designed and implemented a scalable, metadata-driven ETL framework using PySpark, Spark SQL, and Azure Data Factory; owned end-to-end data ingestion, transformation, orchestration, and cloud data architecture. - Impact: delivered a reusable, cloud-native data pipelines that improved data quality, governance, and scalability.

Diverse AI and Data Engineering experience across insurance, retail, and R&D. • Data Scientist – Prospectus Distribution Optimization | Intermarché Designed a data-driven optimization model to improve prospectus distribution efficiency across diverse communities. Analyzed historical distribution data and community-level features to determine the optimal prospectus type and volume per location. - Tech Stack: Azure Databricks, Supervised Learning, Unsupervised Learning, MLFlow • Data Scientist | Document Recognition | AXA Developed and trained AI-powered document reading models for 11 insurance providers, including driver licenses, insurance history, etc. Built hybrid pipelines combining Azure service with custom post-processing. Delivered improved extraction accuracy by 40%. - Tech Stack: Azure Form Recognizer, Azure Machine Learning, Python, Azure Cognitive Service • Data Engineer | R&D: Cloud-Edge Digital Twin Analytics Platform | Accenture Implemented the edge-layer Digital Twin solution to enable real-time operational insights for simulated military scenarios. Built and deployed Digital Twin, designed graph data models in Neo4j, and developed APIs to query and expose twin state. - Tech Stack: Digital Twin, Neo4j, Azure Functions, Docker, Azure Machine Learning

Extensive PoC experience with a focus on recommendation and NLP. • Data Scientist | Graph-Based Recommendation | Engie Designed and delivered a graph-powered recommendation system to surface the most relevant articles. Built a Neo4j graph database and exposed it via a GraphQL API (Apollo Server) for low-latency, flexible querying. Integrated NLP techniques and automated web scraping (Goose, Newspaper) to enrich content metadata, improving recommendation relevance and search precision. Tech Stack: Neo4j, GraphQL, Azure Cognitive Service, NLP, Apollo, Goose, Newspaper, Web Scraping, Cypher • Data Scientist | PII Obfuscation Led the benchmarking of diverse NLP methodologies to automate PII detection in medical certifications, ensuring GDPR compliance. The modular PII solution that was adopted as the internal reference standard for multiple PII-focused projects. - Tech Stack: Python, spaCy, Stanza, Azure Cognitive Services, OCR, Pytest, Azure DevOps • Data Scientist | Enterprise Discovery & Recommendation Engine Developed a POC recommendation engine to optimize the discoverability of internal dashboards, documentations, datasets across global business units. Benchmarked statistical, knowledge-based, and collaborative filtering models. - Tech Stack: Python, Collaborative Filtering, Azure SQL, SSMS, Data Modeling • Data Scientist | Crisis Communication & Sentiment Clustering Built an analytics application to monitor negative public sentiment on Twitter (X), delivered actionable insights on emerging negative trends, supporting proactive brand reputation management. - Tech Stack: Python, NLP, scraping, data cleaning, sentiment analysis, NER, keywords extraction, FastText, Clustering, Dash, Twitter API, SQL, data visualization

Product Manager for digital transformation at Shanghai Media Group, leading digital strategies and roadmap for 10 TV channels to improve audience engagement. - Developed and implemented online video editorial strategies. - Developed and implemented social media promotion strategies. - Integration of TV broadcast content into the official website and mobile applications. - Stakeholder communication and cross-functional collaboration.

Journalist and Editor of multiple English-languague talk shows of Shanghai Media Group. - Responsible for topic selection, scriptwriting, question preparation, and post-production editing to ensure engaging and high-quality content; - Coordinated with stakeholders, including policymakers and industry experts, to facilitate in-depth discussions on current events.