Deconstructing Conversational AI - Design and Best Practices

Jatin Gupta (~jatin56)


Conversational AI, or Voice Assistant (VA), is essentially changing the way we interact with our users. Even though VA is not exactly new to the technological sector and speech recognition has been in use for almost 20 years, it has seen an exponential increase in adoption with the onset of the pandemic - with many firms across the world deploying enterprise chatbot solutions on platforms such as DialogFlow, Watson Assistant, etc. to provide customer care services, vaccination appointments and much more to their customers spread across different regions, cultures, languages, and preferences. Even in situations when these solutions are not able to provide end to end solutions, they are being used to assist customer care executives with document search, collecting feedback, and sentiment analysis.

Now, that companies are realizing the time and cost benefits of deploying VAs and with the uncertainties of the pandemic, there are only going to be more VAs. At IBM, I am working closely with one of the largest retail chains in USA and helped them build a complex (yet simple) VA from the ground up. I hope to use the lessons I learned from that experience to educate Data Science practitioners about the following topics (not exhaustive):

  1. The history of VAs and why they are relevant
  2. How to design and build user-centered VAs
  3. How Voice is different from Visual and Text
  4. How to build Multi-Modal and Multi-Channel VAs
  5. VA development life cycle - components and integration
  6. Best Practices and Common Mistakes



Speaker Info:

Jatin Gupta currently works as a Data Scientist at IBM. Having completed his graduation in Computer Science and Engineering, he loves working with new technologies and collaborating with people to solve today’s problems and his day-to-day involves him working with technologies such as AI, ML, NLP, IBM Watson, etc. Having always been deeply fascinated by International Politics, Jatin has keen interest in how technology affects today’s politics and how it can be used for the ‘greater good’. He also likes to explore the role of ethics and design in Artificial Intelligence and delivered a talk on ‘Ethical Considerations in AI’ at BelPy 2021 and on 'Design Thinking in AI' at Python Web Conference 2021. In addition to that, he also contributed a chapter on 'Impact of Technology and the Internet on International Politics' in the book Novus Politika. When he is not busy writing code on his laptop, Jatin spends most of his time reading, writing, traveling the world, watching movies, and annoying the people around him. An admitted football fanatic, he feeds his addiction to football by watching Premier League games on a Sunday afternoon. Jatin tries his best to give back to the society and offers pro bono career coaching to college students. He has also worked as a teacher volunteer with NGOs such as Vishvas and ActForChildren and is presently working as a mentor with MentorTogether.

Speaker Links:

My previous talks have not yet been publicly published so I am attaching my linkedin profile URL in case you want to know more -

Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Beginner
Last Updated: