Dina Baddaje

Building AI agents using generative AI (ChatGPT, Gemini, Mistral, etc.), RAG (Retrieval-Augmented Generation), and Java/Groovy microservices. - Developing agents to generate SWIFT banking messages from customer emails: extraction, analysis, validation, routing, email sending, and drafting of SWIFT messages. - Extracting KPIs from non-standardized PDF documents in more than 10 languages into SQL tables. - Designing automated workflows for multi-format document ingestion and targeted document searches. - Deploying and operating AI pipelines in production, integrating via APIs, and working in an Agile environment with IT teams. - Designing testing methodologies and ensuring non-regression through Python, Ragas, and evaluation-agent tests.

Leverage LLMs and data science methodologies to design, evaluate, and optimize AI-powered customer service solutions at scale. - Analyze over 1 million conversations across 7 industries (insurance, banking, mobility, marketplaces, etc.) to identify high-ROI workflows for automation using LLMs. - Design and implement a batch LLM inference framework for secure, large-scale internal data processing. - Apply Retrieval-Augmented Generation (RAG) to ensure contextual relevance and factual consistency in FAQ-style responses. - Perform unsupervised clustering using LLM embeddings to identify recurring agent behaviors and frequent support intents. - Build LLM-powered anonymization pipelines to handle sensitive customer data across highly regulated sectors. - Benchmark LLM-based pipelines against traditional NLP and NER approaches, using custom KPIs to assess quality and accuracy. -Lead audit initiatives on multi-million record datasets to evaluate and improve both AI and human agent performance.

In charge of data needs within the Identity Verification Team (formerly Ubble) - Designed and implemented an accuracy measurement protocol based on a Bayesian sampling method with an optimized alert threshold. - Developed real-time performance dashboards to monitor an automated identity verification solution (human-free intervention). - Maintained and enhanced the analytics platform (DBT, Snowflake) to ensure data reliability and reporting accuracy. - Provided support to internal stakeholders and clients by addressing ad-hoc data analysis and reporting requests. - Developed a technical chatbot for sales teams using RAG architecture with Gemini during a Hackathon.

- Build business KPIs and measure performance across products and teams. - Develop and deploy reporting and monitoring products. - Design and implement tracking plans to map user flows. - Strengthen the data model by improving processing and aggregations. - Set up and train a data ambassador program across all verticals. - Lead a revamp project on the BI tool (accumulating +18k graphs since 2019).

Reporting: - Development of a reporting and dashboarding tool of an automotive company sales data . - Creating data aggregation functions, KPI comparision dashboards and sales previson tables. Marketing Mix Modelling: - Analysing factors that impact a brand sales (marketing actions, competitors, stocks, launch of products...) - Data tracking and processing. - Modelling using econometrics techniques. Segmentation: - Segmentation of an insurance company clients portfolio based on the client contracts and values.