Lead AI Engineer

We are building a competitive intelligence & insights platform for marketing teams. We analyse paid media dynamics across social platforms (Meta, TikTok, etc.) to provide a real time quantitative vision of the market media strategies, and generate media planning recommendations.

We already serve multiple paying clients across various industries, and are in process of closing our seed round to help us scale. We are now hiring an A-team in Marseille that will consolidate the foundations to enable the scale.

Our tech stack revolves around the following pillars:

• Data collection — gathering ads (metadata, media (image, video, text), audience volumes, targeting, etc.) & third party data (product taxonomies, etc.). It needs to be exhaustive, in real-time, and highly scalable.
• Data transformation — turning fragmented platform-specific ad data into a unified, cross-platform model that humans, machines, and AI can query and compare without friction.
• Multimodal AI — identifying brands, pinpointing products & matching to clients' taxonomy, identifying creative & media elements (media objective, influencer identification, etc), generating insights & recommendations from data, etc. Needs to be reliable, fast & cheap.
• Intuitive and actionable product — for our customers to interact with the competitive data: analytics dashboards and ad search, API / MCP, natural language query interface.

Stack

We try to stay simple & pragmatic:

• Infra — Everything runs on Google Cloud: Cloud SQL (Postgres), BigQuery, Cloud Run (services & jobs), Vertex AI. All workloads are containerized. Terraform is on the roadmap.
• Backend — Python 3.11+, FastAPI, Pydantic, SQLAlchemy. Typed end-to-end, no ORM magic. Alembic for migrations, Postgres for storage, Cloud Run for compute.
• Frontend — Vue 3 + TypeScript + Vite. Nuxt UI for components, ECharts for visualization. Playwright for E2E.
• Data & Analytics — dbt for transforms & data tests, BigQuery as the datawarehouse & Omni Analytics for analytics & visualization. Clean separation between operational and analytical workloads.
• AI — Off-the-shelf LLMs (Anthropic, OpenAI, Google) with fine-tuning when needed. Embeddings for semantic search, RAG for retrieval, structured outputs for automation, evals to keep quality tight. Simple & pragmatic.
• DevX — Claude Code is central to how we ship code. Custom agents, skills, and internal tooling to stay fast — we keep pushing as the capabilities mature.

The role

We are looking for a Lead AI Engineer. You will join as one of the first engineering hires, working directly with our co-founder & current tech lead Charles, and our soon-to-join full-stack Tech Lead.

Our AI layer is the backbone of our product. It runs everywhere: brand & product identification, creative & media feature extraction, insight generation, similarity & campaign detection, natural-language querying, etc. Even though the foundations of the AI layer are already in place, your mission will be to make this layer scalable.

What you'll work on

Architect & design the AI layer

• Create end-to-end architectures for our AI systems: ingestion of multimodal ad creatives, embedding & indexing pipelines, retrieval, LLM-based extraction and generation, structured outputs, agentic and natural-language interfaces.
• Implement proper traceability for all our inferences, tracking cost, quality & latency, and enabling continuous measurement & optimisation.
• Design solution components: evaluation infrastructure, cost & token observability, tracing, model routing, caching layers, batch & async pipelines, safe rollout mechanisms (shadow, canary, A/B).
• Ensure solutions meet our requirements for quality, latency, cost, resilience, and maintainability. Bring SRE thinking to AI: SLIs/SLOs on model quality and pipeline health, incident response, error budgets, production monitoring.
• Contribute directly to building the AI layer: writing design docs and architecture, writing production code (Python, FastAPI, Pydantic, async pipelines), prototyping and benchmarking models, embeddings, prompts, validating technical approaches with real data, and maintaining and extending our CI/CD, evals, and AI observability stack.
• Support the team during complex engineering tasks, unblock challenges, and make sure what ships matches the architecture.
• Balance high-level architectural thinking with pragmatic, hands-on execution. You're as comfortable writing a design doc as you are debugging a prompt at 11pm.

Guide delivery & technical leadership

• Co-define the AI roadmap with the founders: translate product vision into delivery plans, prioritize ruthlessly, make build-vs-buy-vs-delete calls.
• As the team grows, le