At Second Nature, we build AI-powered training experiences that simulate real-world conversations using speech and generative AI. Conversation quality is at the core of our product — how realistic, effective, and consistent an interaction feels directly impacts how well people learn.

We’re looking for an NLP Data Scientist with deep expertise in conversational AI to help define, evaluate, and continuously improve the behavior of our AI systems. This role is highly product-embedded, working closely with product managers, Applied AI Engineers, and customer-facing teams to deliver the best possible conversation experiences.

What You’ll Do

• Analyze and evaluate AI-generated conversations across a wide range of real-world training scenarios
• Define and evolve what “good” looks like for conversational quality, realism, and effectiveness
• Design and run qualitative and quantitative evaluation frameworks for conversational AI systems
• Lead systematic prompt analysis and experimentation to improve conversation behavior at scale
• Work hands-on with Python and AI systems to prototype, integrate, and validate improvements
• Leverage feedback from a dedicated group of conversation experts and testers to identify gaps and opportunities
• Partner closely with product, customer teams, and Applied AI Engineers to translate insights into product and system changes
• Use insights from thousands of unique conversations generated daily to guide system and feature improvements

You’ll Be a Great Fit if You Have:

• A M.Sc. / Ph.D. in Computer Science, Machine Learning, or a related field, or equivalent practical experience
• 3+ years of experience as a Data Scientist in non-academic settings, ideally with a focus on Large Language Models, NLP, or Speech Processing
• Experience working deeply with NLP or conversational AI systems
• Strong intuition for language, dialogue, and conversational dynamics
• Experience evaluating or improving generative language systems in production
• Ability to design experiments and evaluation criteria for complex language behaviors
• Fluency in Python and comfort working with data, experiments, and analysis workflows
• A collaborative mindset and ability to communicate insights to technical and non-technical partners

It’s a Plus if You Have:

• Experience with conversational evaluation methods (rubrics, human review, model-assisted evaluation)
• Experience with LLMs, prompt design, and conversation analysis
• Background in dialogue systems, computational linguistics, or applied NLP research
• Experience with speech-based or multimodal conversational systems
• Familiarity with AI training, education, or role-play simulation contexts

Why Join Second Nature

• Shape the quality and realism of AI-driven conversations used by real learners every day
• Work on a product used in production daily by Fortune 500 and Fortune 50 companies
• Work with rich, high-volume conversational data from production systems
• Influence product direction through deep understanding of conversational behavior
• Apply NLP expertise to problems where nuance, tone, and realism truly matter
• Be part of a small, thoughtful team pushing the boundaries of conversational AI