toptal - AI Researcher
Requirements
• PhD in Computer Science, Machine Learning, AI, Electrical Engineering, or a related field. • 5+ years of experience in applied AI research or ML systems with production impact. • Strong background in large-scale machine learning, LLMs, or multimodal AI systems. • Fine-tuning large language models. • Reinforcement learning methods (RLHF, DPO, or GRPO-style approaches). • Strong understanding of representation learning, embeddings, and joint embedding spaces. • Experience with speech and audio modeling, including STT, ASR, or audio signal processing. • Proficiency in Python and modern ML frameworks (PyTorch, Hugging Face ecosystem). • Experience designing or improving evaluation methodologies for LLMs or agentic systems. • Experience with agentic AI systems, including reasoning, planning, or tool-use architectures. • Background in multimodal AI systems (text, audio, vision, or structured logs). • Experience embedding AI into real-world products (browsers, IDEs, enterprise tools). • Experience with real-time or streaming AI systems. • Open-source contributions or publications in top-tier ML/AI conferences. • Strong ability to define research hypotheses from ambiguous, real-world problems. • Outstanding written and verbal communication skills in English. • You must be a world-class individual contributor to thrive at Toptal. You will not be here just to tell other people what to do.
Responsibilities
• Advance research on agentic AI systems trained on real-world interaction signals and multimodal data. • Design and experiment with learning paradigms for large-scale models, including RAG, supervised fine-tuning, RLHF, DPO, and GRPO-style methods. • Develop multimodal representation learning approaches, including joint embedding spaces across text, audio, logs, and structured interaction traces. • Improve speech and audio intelligence capabilities, including STT, ASR, and audio-driven learning signals. • Research methods for enhancing agent reasoning, planning, tool use, and adaptation in real-world environments. • Define how complex behavioral and interaction signals can be translated into effective training objectives for large-scale models. • Build and refine evaluation methodologies for agent performance in real-world, domain-specific scenarios. • Collaborate with engineering and product teams to bring research ideas into production systems. • Identify patterns in real-world workflows and convert them into generalizable modeling and representation strategies. • Contribute to the long-term research direction of Toptal’s agentic AI systems and multimodal capabilities. • Stay current with academic and industry research and integrate relevant advancements into internal systems. • In the first week, expect to: • Access our existing datasets, agent stacks, and internal evaluation tools. • Map the landscape of raw data sources currently feeding our agentic systems. • In the first month, expect to: • Develop a deep understanding of our current architectures and evaluation methodologies. • Identify high-leverage gaps where data improvements can measurably increase agent capability. • Initiate concrete improvements to pipelines converting raw inputs into model-ready assets. • Shape feedback loops that utilize live performance as a training signal. • In the first three months, expect to: • Own a production data pipeline from ingestion through delivery into RL or fine-tuning workflows. • Define reusable schemas that abstract repeated workflows into queryable formats. • Drive measurable advancements in agent accuracy within a specific vertical, backed by metrics. • Integrate AI features into user-facing surfaces like browsers or enterprise tools. • In the first six months, expect to: • Lead the design of multimodal pipelines that unify text and real-time logs for agents. • Establish tooling for encoding institutional knowledge into scalable schemas for the team. • Define the team’s strategy for fine-tuning and capturing human feedback for RLHF. • Mentor teammates on data-centric approaches and influence the team’s technical direction. • In the first year, expect to: • Serve as a key technical leader in turning proprietary data into a durable competitive advantage. • Operate as a recognized expert across the team on knowledge representation and improvement loops. • Drive a step-change in agent capability across multiple verticals through clear performance metrics. • Shape the next generation of products by evolving data, agents, and applications together.
Apply in one click
Upload My Resume
Drop here or click to browse · Tap to choose · PDF, DOCX, DOC, RTF, TXT