Research Scientist

Summary

Join Canvs AI, a company using AI to understand consumer emotions, as a Research Scientist. Reporting to the Director of Data Science & AI, you will develop next-generation AI systems transforming how organizations understand customer emotions and loyalty. You will lead the transformation to transformer-based emotion classification, design and implement various LLM agents, develop evaluation frameworks, create benchmarking systems, and research novel approaches to emotion detection. Collaboration with engineering teams and presentation of findings to product teams are key aspects of the role. This position requires a strong background in transformer models, text classification, and LLMs, along with excellent communication skills.

Requirements

• MS or PhD in quantitative field (computer science, economics, ecology, math, physics, psychology, quantitative social science, etc.) with 3+ years of industry experience
• Strong expertise in transformer models and text classification systems
• Experience with large language models and retrieval-augmented generation (RAG)
• Proven track record implementing and evaluating NLP models in production environments
• Strong software engineering skills with Python and modern LLM/ML frameworks
• Experience with model monitoring, validation metrics, and quality assurance for AI systems
• Skilled at working with unbalanced datasets and handling taxonomy mapping challenges
• Self-directed and able to prioritize effectively in a fast-paced environment
• Passionate about using AI to understand human emotions and drive customer insights
• Excellent communication skills and ability to explain complex technical concepts
• Experience working with remote or distributed engineering teams
• Ability to perform all the above according to applicable security and privacy best practices while staying up to date with ever-evolving standards in the industry

Responsibilities

• Lead the transformation from keyword-based to transformer-based emotion classification systems that process millions of texts daily
• Design and implement a variety of LLM agents that provide accurate, trustworthy insights. Examples include: RAG-based LLMs, fine-tuned open source models, SQL lookup agents, visualization agents, statistical agents, and more
• Develop evaluation frameworks to detect model issues before they reach customers
• Create benchmarking systems and validation datasets for various text-classification analysis models as part of our AsaCore initiative
• Research and implement novel approaches to emotion detection that can handle nuanced customer feedback
• Collaborate with offshore engineering teams to deploy models to production
• Present findings to product and go-to-market teams to inform product roadmap decisions
• Stay current with the latest advancements in NLP, transformer models, and LLM applications

Preferred Qualifications

Experience reducing hallucinations in LLMs through techniques like RLHF, policy models, constitutional AI, or other guardrail implementations

Benefits

• Salary Range: Commensurate with experience and title
• Tremendous opportunity for career growth, and performance bonuses
• Strong culture and commitment to values
• Flexibility via a fully distributed company (Work From Anywhere!)
• Generous health benefits
• Unlimited paid vacation
• Summer Fridays
• Monthly (at a minimum) virtual team-building events