YAML Configuration Editor
Custom internal web app that allows non-technical stakeholders at papernest to safely configure data pipelines. Cut pipeline build & test time by ~50% and incident investigation from 30–60 min to under 2 min.
Context
At papernest, data pipeline configuration relied on manually editing YAML files distributed across multiple markets. Non-technical stakeholders — Sales Ops managers and Data Analysts in FR, ES, and IT — had no safe way to adjust configurations without involving a Data Engineer. A single YAML syntax error could silently break pipelines affecting multiple markets.
Approach
- ·Designed a Next.js/React web interface with real-time YAML validation and an intuitive form-based editor
- ·Built a Python backend handling API calls with the internal ETL engine and LLMs
- ·Created a context-aware UI translating YAML structure into understandable fields for non-technical users
- ·Integrated an AI Copilot assistant guiding stakeholders through creating, editing, and repairing YAML configurations
- ·Implemented a testing framework to preview and validate business logic before any production deployment
Solution
The editor gives non-technical stakeholders full control over pipeline configuration through a guided, safe interface. All changes are validated in real-time, previewed before application, and deployed through a controlled flow — with no risk of introducing syntax errors into production pipelines.
Key outcome
Pipeline build & test time cut by ~50%. Incident investigation went from 30–60 min to under 2 min. Sales Ops teams across FR, ES & IT now manage configurations autonomously, freeing engineering time for higher-value work.
Applications
- ·Multi-market data pipeline self-service
- ·Non-technical stakeholder autonomy
- ·Reducing engineering bottlenecks in data operations
- ·Safe configuration management
Project details
Technologies
More projects
View all →Log Data Visualization Platform
Full-stack internal web application for log data visualization and analysis at Bouygues Telecom. Machine log investigation time reduced from several minutes to near-instant access. Designed and built independently from scratch.
SQL Streaming Module – Documentation & Evolutions
Contributions to an internal SQL-on-Flink streaming module at the core of Bouygues Telecom's ETL. Main deliverable: an automated documentation system and CI/CD integration to make the module production-ready.
EscobAddictions — Global Drug Data Analysis
Exploratory data analysis on global drug phenomena across four independent studies: mortality by substance, consumption by age group, cocaine–crime correlation, and drug-related imprisonment trends. Includes an interactive Power BI dashboard.