Arthur Gallouin

Lead engineer for Qobra’s data ingestion platform, processing 100M+ rows/day across 6+ heterogeneous sources (CRMs, data warehouses, flat files, public API). Responsible for performance, reliability and architectural direction of all ingestion flows.

• Reduced full-import duration from 2h17 → 14 minutes by redesigning reference propagation, batching logic, and thread-pool isolation (eliminating cross-tenant blocking).

• Improved ingestion reliability with nightly full resyncs, reference-consistency scanners and regression tests, reducing customer-visible reference issues from 3/week → ~1/month.

• Stabilized differential syncs by addressing computed-field drift caused by unreliable update timestamps in upstream CRMs.

• Built tooling and dedicated test environments for developers & Customer Success, increasing debugging speed and autonomy.

• Introduced the Shape-Up methodology (6-week cycles), strengthening prioritization, delivery consistency and cross-team alignment.

• Ensured schema consistency and correctness for the Public API, consumed by multiple engineering teams.

Built analytics features turning conversations/tickets into actionable metrics (SLA compliance, response times, workload distribution) and improved the foundation of the Insights data model.

• Led the migration from heuristic “conversation stages” to an explicit ticket-resolution workflow, improving data correctness and simplifying downstream analytics.

• Designed a full local Elasticsearch environment with synthetic dataset generators, drastically improving developer experience and feedback loops.

• Reintroduced regression tests for analytics metrics, clarified metric specifications and corrected long-standing inconsistencies.

• Delivered full-stack contributions across backend Elasticsearch/Painless pipelines and frontend dashboard components.

Scaled the engineering team from 3 → 15 developers; defined team topology, recruited engineers, and established architectural and coding standards.

• Led the migration to a CQRS + event-sourcing architecture (Kafka + Postgres), improving scalability, decoupling and long-term maintainability.

• Reduced lead time to production by 50% through CI/CD standardization, improved review workflows and a unified design system (200+ UI components → 60).

• Introduced clean architecture and domain-driven boundaries enabling autonomous cross-functional feature teams.

• Acted as technical owner across squads, ensuring product/engineering alignment for Skillup’s HRIS SaaS platform.