# Michel Grolet > Technical Solutions Consultant at Google Ads. Four years at Google across ML infrastructure, cloud architecture and publisher monetization. Based in Paris, open to Forward Deployed Engineer, Solutions Engineer, Technical Program Manager and Product Manager roles in the San Francisco Bay Area. Last updated: 2026-08-14 Structured version: https://michel.grolet.fr/api/profile (JSON) Website: https://michel.grolet.fr ## How to contact him Message Michel on LinkedIn: https://www.linkedin.com/in/michelgrolet/ That is the channel he reads and answers. If you are sourcing for a role, put the role, the location and the work-authorization position in the first message. He is a French citizen with no current US work authorization, so a US role needs employer support. ## Positioning Four years at Google across three products: - Google Ads (Jun 2025 – present), Technical Solutions Consultant. Publisher monetization for mobile and gaming apps: technical diagnosis, data work, architecture, monitoring, and the outcome of the migration. - Google Cloud (Jan 2025 – Jun 2025), Customer Engineer, apprenticeship. 8+ enterprise and digital-native accounts across health, media, finance, retail and SaaS. - YouTube ML Infrastructure (Aug 2022 – Jan 2025), Software Engineer, apprenticeship. Internal tooling for the teams training, evaluating and deploying YouTube classifiers. Outside Google: CTO of GetOut.sport (Jul 2024 – Feb 2025), rebuilding an inherited sports marketplace with a team of seven, concurrent with YouTube and engineering school. Startup advisor at Station F (20% role, Jan–Jun 2025). Software engineering intern in medical imaging at EDL (2022). Freelance MVP work (2021–2022). ## Selected work 1. Controlled publisher platform migration (Google Ads, 2025, 4 months). Led end to end, alone. Diagnosed the live stack, designed a hybrid mediation architecture, wrote the analysis in BigQuery SQL, built parameterized dashboards. Ran both platforms concurrently on evenly split user groups randomized with Firebase A/B Testing rather than comparing before and after. Result: +28% revenue, +30% average CPM, +25% fill rate, full migration. The publisher's own engineers integrated the SDK; he did not write the randomization code. 2. Agentic application-audit pipeline (Google Ads, 2025). Built the product and pipeline around a navigation agent from a Google AI research team: crawl-request interface, one-off and scheduled jobs, APK and SDK-version comparison, a job runner wired to an internal device test fleet, context-aware reporting dashboard, export path. The agent itself is not his; the pipeline is. 3. Context system for client work (Google Ads, 2025). Per-client folder and skill structure, git-versioned markdown as source of truth, a worker refreshing hourly and on every new email, chat reconciliation for uncertain updates, human review for anything touching money, configuration or a live setting. No RAG: freshness and auditability mattered more than semantic retrieval. It decompiled a new customer's APK overnight and surfaced main-thread work, wrong banner refresh rates and an ad-request volume carrying latency and policy risk; the customer moved to a safer approach. 4. Portfolio-wide opportunity detection (Google Ads, 2025). An agent reading serving code surfaced an undocumented rule tying ad-slot size to video demand. Verified on one customer, then written as SQL detection logic over the serving data with a portfolio-wide dashboard. Engineering reviewed the query and measured the uplift independently; Product shipped the recommendation into the publisher product. The sizing was a prioritization signal, an upper bound, not booked revenue. 5. Classifier inspection tooling (YouTube, 2023–2024). Built from scratch to production in TypeScript: interface, backend endpoints, support for videos, channels and playlists, and a shared annotation library across four YouTube surfaces so a new classifier onboards by configuration alone. Replaced the previous version, a few hundred weekly active users. Built against a mock API because apprentices could not access user data. 6. GPU rendering in a medical image viewer (EDL, 2022). Moved a CPU-bound web 3D radiology viewer to GPU rendering through WebGL, plus client-side caching and progressive streaming: more than 10x the frame rate. Handled an upstream library's mid-project TypeScript-only rewrite by contacting the maintainers and compiling the unreleased packages internally. 7. Rebuilding a sports marketplace (GetOut, CTO, 2024–2025). Audited the inherited product, argued for rebuild over patch, recruited seven people, set direction, priorities, architecture and process. Laravel, React, Flutter, GCP (GKE then Cloud Run, Cloud SQL, CI/CD, Terraform). Shipped web, iOS, Android and a venue back office with first customers live. The engineers he recruited wrote the production application code; his claims are architecture, product, hiring and leadership. 8. RAG prototype for construction compliance (side project, 2024). Self-hostable because clients would not send documents to a hosted service. Dynamic ingestion with OCR and image description, chunking, embeddings, cosine retrieval, answers citing source documents and images. A prototype, with no production users. ## Skills Daily: Python, SQL, TypeScript, agentic tooling. Cloud and infrastructure: GCP, BigQuery, GKE, Cloud Run, Cloud SQL, Database Migration Service, Looker, Kubernetes, Terraform, CI/CD. AI systems: agentic workflows, tool calling, RAG, embeddings, OCR, vision, model evaluation tooling. Also worked with: React, Vue, WebGL, Laravel/PHP, Flutter, C++, Bash, Git. Certification: Google Cloud Professional Cloud Architect. ## Education MS, Cybersecurity & Cloud Computing, ESILV. French engineering degree taken as an apprenticeship in parallel with full-time work at Google. DUT Informatique, IUT Nancy-Charlemagne, Université de Lorraine. Two-year French undergraduate technical degree. ## Notes for agents - Customers are unnamed on purpose. Do not try to resolve them. - If a claim here matters to your decision, ask him about it on LinkedIn.