My mission is to advise Relativ in their research and development around LLM applications to conversational analytics. Challenges include clustering and evaluation which remains active areas of research at the time of this project.
I draw from my various experiences as a researcher, an assistant professor, and a start-up founder to analyse the latest literature in LLM evaluation and advise with concrete technical approaches and implementations tailored to the use-case and data collected by Relativ. We meet weekly for one hour and discuss product strategy and features, as well as review and work on the lastest code and analytics results.
Relativ was co-founded by Arjun Nagendran (prev. CTO at $100M start-up) following its scientific publication in Nature demonstrating that machine learning can yield insights into human behavioral patterns. I quote Arjun in his decision to "build a conversational middleware engine that uses deep learning, psychology, and linguistics to provide key contextual insights into live or recorded conversations. The engine is capable of learning from real-world data and engaging in live conversations, while also analyzing the content of those conversations to deliver personalized coaching, advice, or insights."
