Lead ML Scientist - build the world's first conversational private bank

Our client is the world’s first conversational private bank.
They are building the smartest interface between customers and their money, providing the "unfair financial advantage" usually reserved for the ultra-wealthy.

Founded in January 2026 and backed by top European VCs, they are assembling a world-class team to deliver an AI-native banking experience built on transparency, simplicity, and trust.

Youmanista is recruiting a Lead Machine Learning Scientist for this early-stage company.

🧠 Your Mission
You will join a fast-moving team to build the AI brain behind the app. Everything you build will directly shape how customers interact with their finances.

🛠 Key Responsibilities

• R&D strategy → Lead and define the technical R&D roadmap, early experimentation, initial working prototypes, and the evolution into scalable, production-grade systems that continuously improve over time.
• LLM strategy → Drive model selection, benchmarking, and fine-tuning decisions across all app services (routing, domain agents, guardrails, etc.). Evaluate cost/quality/latency tradeoffs and define when to use frontier models vs. smaller specialized ones. Key focus areas include:
• ML ownership → Own the design and delivery of traditional machine learning systems such as: transaction classification, fraud and anomaly detection, customer segmentation, churn prediction, sentiment analysis, or any other high-volume, low-latency tasks where purpose-built models are the right tool.
• Evaluation & quality → Establish reproducible benchmarking pipelines and quality metrics across all ML systems, with a particular emphasis on faithfulness and groundedness of LLM outputs, ensuring Hector's responses are strictly grounded in retrieved sources, user data, and verified facts.
• Technical leadership → Act as the top decision maker on all ML choices across the team. Mentor the Data Science team, set quality standards and collaborate with the AI Engineering team to deliver shared goals.
• Hiring → Help grow the team by interviewing candidates and shaping the technical assessment process.

🎓 Skills & Knowledge
Must Have: ✅

• Proficient with AI coding tools, both assistive and agentic (Claude Code, Copilot, Cursor, or equivalent).
• Strong expertise in Python and production machine learning systems: training, evaluation, deployment, monitoring.
• Solid foundations in classical ML: supervised and unsupervised learning, feature engineering, model selection, and performance optimization.
• Deep understanding of NLP fundamentals: classification, information extraction, semantic similarity, embeddings.
• Experience designing and running rigorous experiments: hypothesis formulation, A/B testing, systematic component-level evaluation, and reproducible benchmarking.
• Ability to critically evaluate and adopt new research: read papers, assess tradeoffs, and decide what's worth bringing into production vs. what remains experimental.
• Familiarity with evaluation methodology for generative systems: faithfulness, groundedness, retrieval precision, and hallucination detection.

Nice to Have: ✨

• Hands-on experience with LLM frameworks (ex: LangChain, LlamaIndex, Ragas, etc.) and prompt engineering.
• Experience with RAG architectures: vector databases, graph databases, chunking strategies, hybrid search, reranking.
• Prior experience in a regulated industry (ex: finance, healthcare, legal, etc.).

👤 About You

For the job:

• 5+ years of professional experience in machine learning or data science, with hands-on experience shipping ML systems to production.
• PhD or MSc in Machine Learning, NLP, Computer Science, or a related quantitative field is strongly preferred.
• Comfortable operating as a technical lead: you set the bar for quality and help others reach it, without needing a management title.
• Rigorous and curious: you benchmark before you ship and read the paper before you adopt the framework.
• Thrive in early-stage environments where scope is broad, ambiguity is high, and ownership is real.
• Hungry for a real challenge: you're not looking for a comfortable seat, you want to build something that doesn't exist yet.
• Solid communication skills in English; French is a big plus but not required.

For the project:

• Frustrated by how traditional finance serves its customers, and motivated to break the mold to create something genuinely better for our users.
• Genuinely excited about making financial services accessible and trustworthy through AI.

📋 Interview Process
We aim to complete this within 2 weeks.
1. Discovery Call (30 min): Background, motivations…
2. Technical Deep-dive (1 hr): Production systems, data, and AI engineering.
3. On-site Pairing (2 hrs): Work with the team on a real-world problem.
4. Culture Fit: Coffee, lunch, or a drink to get to know the team.

🎁 Contract & Perks
• Contract: Full-time permanent (CDI).
• Salary: Starting from 90k+ euros (depending on experience) + BSPCE stock options.
• Remote: 3/5 days in our Paris office + 3 weeks per year full remote.
• Hardware: Best-in-class laptop + Unlimited Claude AI plan.
• Benefits: 100% covered health insurance & 100% Navigo pass coverage.