Qualifications
You have hands-on experience executing post-training tasks on large language models, such as RFT/RLVR and preference optimization—not merely utilizing APIs or crafting prompts, but actively training and refining models. Your coding skills are robust; the systems you develop must operate reliably, delivering production-ready solutions instead of merely theoretical research outputs. You are proficient in at least one major ML framework (PyTorch, JAX) and have experience with distributed training configurations and GPU clusters. Your understanding of the mathematics behind preference optimization, reward modeling, and alignment techniques is solid, enabling you to determine the most suitable approach for various scenarios. You have either built or significantly contributed to evaluation frameworks that accurately reflect real-world quality, beyond mere benchmark numbers. You possess the ability to diagnose model quality regressions through user-facing issues, analyzing the entire pipeline from serving and inference back to training, and you find joy in this process. You have a strong desire to deliver results quickly and effectively.
About the job
Lovable serves over 2 million users worldwide, helping people build software using clear, direct language. The team seeks an AI Research Engineer specializing in post-training processes to strengthen one of the market’s fastest-growing AI products.
Role overview
This position focuses on the post-training pipeline for Lovable’s AI models. The work blends hands-on model training, production infrastructure, and rapid deployment of optimized models. The team values speed and expects research to become working features in days or weeks, not months. Balancing research with real-world deployment is key.
What you will do
- Oversee the complete post-training pipeline for AI models
- Apply recent research to practical training methods and workflows
- Deploy optimized models quickly to improve user engagement
- Work directly with both model training and production infrastructure
- Deliver results that impact users and product quality
Team and culture
The Stockholm-based team is small, skilled, and values ownership, speed, and collaboration. Lovable’s platform supports everyone from solo founders to global enterprises, enabling rapid transformation of ideas into products. Users span more than 200 countries.
Location
This role is based in Stockholm.
About Lovable
Lovable is revolutionizing software development by allowing anyone to create applications using their own language. Serving a diverse global community of over 2 million users, we facilitate innovation across various sectors, helping individuals and teams transform ideas into reality. Our team in Stockholm is dedicated to fostering an environment that thrives on collaboration and a sense of purpose.
