Paola Ngwe

• E-sensia is a startup developing assistants for emergency medical services (SAMU, SOS Médecin), designed to help operators identify critical calls and automate patient triage using voice and conversational AI.

• Contributed to the development of AI-based medical assistants for call triage and patient pre-selection, bridging raw data and deployable AI systems.

• Designed and implemented data ingestion and processing pipelines for large-scale medical emergency datasets (over 320 tables, some with over 120 columns), using Python, PySpark, AWS, and PostgreSQL within a multi-tiered architecture (Bronze–Silver–Gold).

• Migrated the production database from DynamoDB to PostgreSQL (AWS Aurora). Designed the data model and ensured data consistency across services.

• Developed data quality and compliance monitoring APIs (Flask + PostgreSQL + psycopg2) to validate incoming production data against schema and integrity constraints.

• Created automated matching scripts to link historical audio records with SQL metadata, enabling supervised training of AI models on structured multimodal datasets.

• Developed a natural language data querying tool based on OpenAI models, allowing healthcare professionals to query the PostgreSQL database in English or fluent French, without needing to write SQL code.

• Collaborated with AI researchers to provide clean, labeled, and normalized datasets for training medical call classification and triage models.

• The mission's objective was to provide efficient data solutions for liquidity risk monitoring and regulatory compliance.

• Designed and developed custom tools using Python (Pandas, NumPy, PySpark) and SQL to automate the processing of liquidity risk indicators and financial products.

• Extracted and transformed data from multiple sources, including DataLake, vast internal databases, APIs, and flat files, to deliver tailored results to business analysts.

• Visualized and presented data using Power BI.

• Provided technical support and trained users on internal applications to ensure smooth adoption and operational efficiency.

• Maintained and optimized existing applications, ensuring alignment with business processes and regulatory standards.

• Achieved rapid application development (from 2 days to 1 month) to meet evolving business needs.

• Collaborated with financial analysts and business experts to define and implement data pipelines that met regulatory and strategic objectives.

• Optimized data processing performance for large datasets using PySpark and SQL to improve the efficiency of analytical workflows.

• Worked in an agile environment.

• The mission's objective was to support AXA-IM teams in implementing their cloud migration strategy.

• Automated data ingestion and storage processes from various sources, including on-premises sources, to AXA-IM's centralized data platform, enabling quantitative analysts to access it via a REST API.

• Created pipelines for ingesting, processing, normalizing, and storing daily data in AXA-IM's centralized data platform following a Medallion architecture (Bronze-Silver-Gold).

• Used Azure Data Factory to create the pipelines.

• Developed PySpark tasks for various processing stages.

• Developed processing notebooks on Databricks.

• Implemented CI/CD deployments with Azure DevOps.

• Scrum methodology.