About

Leadership & influence

• Championed and built the platform's observability from scratch — central collector, OpenTelemetry, RED dashboards, alerting and distributed tracing; my recommendation led the org to adopt Grafana Cloud.
• Drove engineering practices org-wide — Dagster→Airflow for large-scale processing, GitOps (ArgoCD + GitHub Actions) with reusable workflows, inner-source, incremental SDLC, TDD and Clean Code/SOLID, and technical writing (docs, RFCs).
• Active in hiring — technical tests, interviews and home-test validation, reading weak signals to tell real candidate work from AI output.

Platform & data infrastructure

• Designed the target pipeline architecture — task-agnostic, scalable, cost-optimized (near-zero cost at idle, auto-scaling for peaks) — Airflow, containerized apps, KubernetesPodOperator and Karpenter; each app independently unit-testable and decoupled from Airflow.
• Defined the AWS networking plan (VPC, CIDR) and a geospatial batch system partitioning daily KPI computation across fields.

AI-augmented engineering

• Use agentic coding tools (Claude Code, Codex, Copilot) for PoCs, IaC/Terraform and review support; use MCP to automate context and ticket/PR creation, and share these practices across the org (AGENTS.md, skills, MCP).

Leadership & people management

• Founded Kpler's MLOps squad — pitched and secured its creation with C-level, proving the value of faster, safer AI/data industrialization.
• Directly managed the squad — 1:1s, OKRs, performance reviews (9-box grid), career development, reporting to leadership.
• Owned end-to-end hiring — skill definition, tests, interviews and validation.

Platform Engineering & DevEx

• Built Kpler's MLOps platform from scratch — self-service training (GPU), deployment and inference for models in prod on AWS EKS with Terraform, GitOps and CI/CD; on Aurora PostgreSQL and Elasticsearch.
• Built a self-service Data platform on Airflow for analysts to industrialize BI/reporting, wiring data marts and KPIs to company stores, monitored and security-compliant.
• Drove DevEx and GitOps — reusable GitHub composite actions & workflows, Helm charts, and an in-house framework of shared modules (helpers, observability, alerting).

Reliability & engineering culture

• Joined the platform-wide on-call rotation — live debugging, fixes and service restoration; created and ran blameless post-mortems.
• Championed observability and "you build it, you run it", bridging R&D with 24/7 operational excellence so data scientists owned their use cases in production.

Leadership & influence

• Managed a Big Data Engineer intern end to end and wrote 360° performance reviews for colleagues I mentored on missions.
• Authored several Quantmetry articles incl. its most-read of 2018 (a Kubernetes overview); wrote chapter 3 of its 5th white paper on MLOps (re)training automation, with a 20-min companion talk.

Platform & data architecture (selected missions)

• Architected layered "medallion" data lakes (raw → cleaned/quality/feature engineering → aggregated business rules) as a recurring blueprint across missions (Kering, EDF, La Sacem, OFI, CFM) on AWS (S3) and GCP (GCS).
• Kering — initiated a group-wide AI Platform-as-a-Service on AWS on a unified S3 data lake serving the group's brands.
• Gustave Roussy — secure multi-tenant AI platform on AWS: equal GPU compute for 20+ teams, clinical data in an HDS-compliant environment locked against exfiltration.
• EDF Renouvelables — 100% serverless, near-real-time IoT platform on AWS for energy-production KPIs at global scale (20 countries, 8,000 wind turbines, 25,000 solar/PV units).

Technical leadership & mentoring

• Paired with and mentored data scientists (incl. the Data Science Manager) and a data engineer — bringing their code up to production standards.
• Drove engineering standards and DevEx — PEP8 quality, CI workflows, configuration & secrets management, packaging/release, Docker, logging/observability; gave the team a Docker stack to run the full Airflow environment locally.

Data engineering & scale

• Industrialized data scientists' PoC code into reliable multi-tenant production with the engineering rigor it lacked.
• Built the daily batch pipeline on self-hosted Airflow (Hortonworks on GCP) with idempotent, self-recovering tasks; ELT across multi-cloud storage and SQL engines (GCS, S3, Redshift, BigQuery) — multi-cloud from 2016.
• Built the real-time stream-processing app (Scala + Spark Streaming) matching desktop and mobile tracker events to an in-house identity-resolution algorithm.

People & technical leadership

• Led and grew a team of 5 engineers — 1:1s, annual reviews, SMART goals, accountability; owned hiring (ran the full loop solo) and people-management cases with HR.
• Interfaced between dev and the wider org (IT, Marketing, Data, Design), aligning priorities into a single roadmap and reporting trade-offs to senior leadership.
• Introduced the organization's first Agile workflows (as JIRA admin, designed a long-cycle project flow with validation gates and a fast-track flow for bugs/blockers).

Performance & scale

• Led the full re-architecture of the product pages — microservices exposing data via REST APIs on a template-engine, SEO-optimized front-end.
• Engineered low-latency rendering under heavy traffic via multi-layer caching (Redis hot data + CDN at the edge) with immediate per-release invalidation of versioned JS/CSS assets.
• Shipped the redesign in Nov 2013; weeks later the surface I owned carried the company's all-time single-day sales record (~€5.2M group GMV at peak holiday load, announced by the CEO).

People & technical leadership

• Managed a team of 4 engineers (annual reviews, growth), steered the product roadmap and owned the architecture decisions across the systems my team built.
• Introduced Redmine company-wide on my own initiative (self-hosted), from leadership pitch to triage and backlog.

Engineering & scale

• Co-built and operated the in-house ad server — engineered for security, high availability, low latency and scale under heavy traffic; delivered premium formats for major brands (LVMH, Sony, Renault, Ubisoft).
• Built real-time pipelines capturing thousands of ad-interaction data points per visitor with zero added latency, powering billing, KPIs, RTB/ad-exchange and dashboards.
• Built from scratch an FFmpeg-based multi-bitrate video encoding platform (5 renditions, 144p–1080p) on a Hadoop-inspired distributed architecture orchestrating AWS instances via the SDK (before Kubernetes).

Engineering

• Front-end & back-end development of multilingual corporate websites for real-estate leaders, plus a collaborative DataRoom SaaS.
• Synced each site's listing catalog with the client's CRM via APIs — removing double data entry while letting sites add exclusive content.
• On my own initiative, structured how we built sites — introduced an ORM, a template engine and AJAX — laying engineering foundations the team lacked; spent a month in Casablanca training the offshore team.

Engineering

• Front-end & back-end development of real-estate web applications far beyond brochure sites — listing search engine, email alerts, saved-criteria client area, admin back-office and CRM integration via APIs.
• Built dynamic, interactive interfaces with JavaScript, CSS and AJAX — updating pages without full reloads for a smoother experience.