Staff AI engineer

BAO; Paris; Hybride; EUR 120 000 - 180 000; Plein temps

Résumé du poste

BAO is seeking a Staff AI Engineer in Paris or London. This founding IC role sits at the center of a tight loop between Curriculum Engineers and Software Engineers to train and improve models in production environments.

You will fine-tune LLMs (SFT/RL), own the data lifecycle from collection to cleaning, and optimize prompts and agent harnesses. The team emphasizes pragmatic shipping, weekly impact, and the chance to help seed a fast-growing startup.

Qualifications

• Real, hands-on fine-tuning experience (SFT/RL) required.
• Strong Python skills and PyTorch experience.
• Experience with the full training data lifecycle.
• Familiarity with prompt/harness optimization (GEPA a plus).
• A pragmatic, shipping-focused mindset in a fast-moving environment.

Responsabilités

• Fine-tune LLMs (SFT and/or RL) for classification and generation tasks.
• Own the full data lifecycle: collection, cleaning, curation, management.
• Optimize prompts and agent harnesses to improve performance.
• Collaborate with Software Engineering to productionize approaches.

Connaissances

Fine-tune LLMs

Python

PyTorch

Data lifecycle management

Prompt optimization

System architecture

Hands-on ML engineering

Outils

PyTorch

LLM tooling

Description du poste

Location: Paris (3rd arrondissement) or London — in-office for the first month, then 3 days/week remote Type: Full-time Compensation: Competitive base salary + equity (BSPCE)

About the company

This is a fast-growing, stealth-stage startup building training environments for the world's leading frontier AI labs. They recreate professional software — think Gmail, Slack, Salesforce, QuickBooks-style tools — pixel-perfectly, to create ultra-realistic environments where AI models learn to use real tools the way a human would.

The traction speaks for itself: $10M in revenue in their first 100 days, a $50M raise closed a few weeks ago, and they're on track for $100M by the end of 2026. They're a lean team (~100 people) split between Paris and London, with plans to hire aggressively over the next 12–18 months.

The role

As an Staff AI Engineer, you'll be a founding IC on their AI team, sitting at the center of a tight loop between two other role families: Curriculum Engineers (who turn software into hundreds of calibrated, verifiable tasks) and Software Engineers (who build the infrastructure that runs everything at scale). Your job is to take the data produced upstream and get the most out of it to actually train and improve the models.

Concretely, you will:

• Fine-tune LLMs (SFT, RL) for classification and generation tasks
• Own the full data lifecycle — collection, cleaning, curation, and management of training data
• Optimize prompts and agent harnesses using methods like GEPA
• Work in tight iteration with the Software Engineering team to productionize and scale the approaches that work

This is a genuinely hands-on, founding role — you're not stepping into something already figured out, you're helping build the foundations. The emphasis is on pragmatism and measurable weekly impact, not six-month research projects.

What we're looking for

• Real, hands-on fine-tuning experience (SFT and/or RL) — this is non-negotiable. API-based LLM usage alone does not meet the bar.
• Strong Python skills, comfort with frameworks like PyTorch
• Experience with the full lifecycle of training data (curation, cleaning, quality control)
• Familiarity with prompt/harness optimization approaches (e.g. GEPA) is a strong plus
• A pragmatic, shipping-oriented mindset — you'd rather have something working and imperfect in production than perfect in a notebook
• Comfort operating in a small, fast-moving, ambiguous environment

Growth path

Multiple tracks are open after this role: Staff/Principal AI Engineer, or a move into technical leadership.

• Based in Paris (3rd arrondissement) or London
• In-office for the first month, then a 3-days-in-office/week hybrid rhythm