Developing NLP features in the industry is different than research in academia. In particular, it doesn’t matter how novel your techniques are unless they have a demonstrable impact on the user and a feasible path to deployment. In this presentation, I will present 3 case studies of how we took NLP research from our ML R&D team and integrated it into our production conversation AI.
For each project, I will describe the problem context, how we identified use cases, technical implementation details, deployment considerations, and factors that led to that project’s success or failure in production.
This presentation is geared toward engineers and researchers interested in the broader perspective, technical managers, and product managers working with NLP or AI teams.
