This is a Full Remote job, available from South America, Europe, and Asia.
About Toptal:
Toptal is a global network of top talent in business, design, and technology that enables companies to scale their teams, on-demand. With $200+ million in annual revenue and team members based around the globe, Toptal is the world’s largest fully remote workforce.
We take the best elements of virtual teams and combine them with a support structure that encourages innovation, social interaction, and fun. We see no borders, move at a fast pace, and are never afraid to break the mold.
Job Summary:
Toptal is building a dedicated AI Research team focused on advancing the frontier of agentic AI systems powered by proprietary real-world interaction data.
We are seeking AI Researchers who are excited to explore how large-scale, real-world signals can be transformed into better reasoning, improved generalization, and more capable multimodal agents.
In this role, you will work at the intersection of model development, multimodal representation learning, and reinforcement learning, designing new approaches that enable agents to learn from complex behavioral data, workflows, and multimodal inputs such as audio, logs, and structured interaction traces.
You will focus on building and improving learning systems for agents, including methods for RAG, fine-tuning, reinforcement learning (RLHF, DPO, GRPO), and joint embedding spaces, as well as speech and audio intelligence capabilities such as STT, ASR, and audio signal modeling.
You will collaborate closely with engineering and product teams to ensure research breakthroughs are translated into scalable systems, and that feedback from production continuously improves model behavior.