In this series we will shed light on the extraordinary community work done by our Rasa Superheroes, take a deep dive into what inspired them, and find out if they have any tips or advice to share with aspiring community builders and conversational AI enthusiasts. In this edition, we get to know__ Yogesh Kothiya!__
Yogesh began his career as a Chip Design and Verification Engineer, but soon after discovering conversational AI, he decided to leave a well paid and comfortable job to start a new journey in the startup space. He was blown away by the potential of chatbots, which led him to take the plunge, and he has never looked back since.
He is now a Rasa Superhero, running local and online workshops, like a recent one in Hyderabad on how to build an AI chatbot using Rasa and the Zomato API. Additionally, he supports workshops led by other hosts in other cities across India and even finds time to run an AI community and consult with chatbot startups there!
Yogesh has a clear passion for conversational AI, and we very much appreciate his help to spread the message about Rasa and empower other AI makers to create AI assistants through his workshops!
Hi Yogesh! What does a typical day look like for you?
I start my day with callisthenics, followed by 10-15 minutes of meditation. I also aim to complete an important task at the start of the day, which is when I feel the most productive. I spend the rest of it building chatbots at work where I am fortunate enough to use Rasa Open Source. After the office, I spend my time on community work which involves lots of planning, management and creating content!
We'd love to learn more about where this all began, how did you find out about Rasa?
If you ask me what made me stick to Rasa, it's the pace at which the company is doing research in conversational AI, making Rasa more powerful, and the amazing community that surrounds and supports it.
!(https://www.datocms-assets.com/30881/1608731788-pyconf-pic-rasa-session3.jpg "Yogesh gave a talk at the 2019 PyConf Hyderabad "AI-Based chatbot building using open source bot-building framework"")
It's always interesting to find out how our contributors took their first step towards becoming a Rasa Superhero, since there are so many ways to get started! What was your first ever contribution to the Rasa open source community?
I always wanted to get involved with open-source projects, and after exploring Rasa and being a Rasa user for a long time, I thought the best way to contribute to the community is sharing knowledge. Therefore, my first ever contribution to the community was a workshop on building Rasa-based chatbots, in Hyderabad, India. Despite it being the first, I felt it was a successful one. Roughly 40 people attended and we had very positive feedback! We did have a few technical difficulties, but managed to resolve those fairly quickly, and we learned to be better prepared in our next workshops (for example preparing and sharing installation guides much sooner in advance). Since then, I have taken workshops in Bangalore, PyConf, Hyderabad and mentored others to host workshops in Delhi, Mumbai, Mathura, and Pune.
!(https://www.datocms-assets.com/30881/1608731792-rasaworkshop-hyderabad-aug-2019-qna.jpeg ""How to build an AI chatbot using Rasa and the Zomato API" workshop Hyderabad, India.")
!(https://www.datocms-assets.com/30881/1608731798-rasaworkshop-bangalore-sept-2019-1.jpeg ""How to build an AI chatbot using Rasa and the Zomato API" workshop in Bangalore, India.")
You did indeed initiate many workshops across India in 2019! Do you have any tips to share with aspiring community builders who would like to grow their local online community?
Yes! The second half of 2019 was really great!
Community builders should always be open for collaboration with other communities who have a similar vision or mission. Always be on the lookout out for passionate contributors who will take the community forward.
A key missing element in some communities is "Engagement" and how many people continue to stick around, I think this is very important when measuring healthy community growth.
You're currently running a Facebook group for conversational AI enthusiasts. Could you tell us a little more about it and how that started?
Well, in my previous domain (Semi-conductor/VLSI) I never came across anything like our communities, and when I switched my career to conversational AI, I quickly realized the importance of open-source communities. Be it through learning from others, sharing knowledge or getting together with others to collectively solve problems, I always felt comfortable enough to share my experiences and this had a huge impact on what I prioritized when building the open-source Co-learning Lounge community.
Could you tell us more about the Co-learning Lounge community?
With Co-learning Lounge, our goal is to come together, to build a collective knowledge pool and open source this education in the tech space. Our prioritizes are providing well-structured, up to date, community created educational content and workshops, personal mentoring, supporting career growth and connections.
Today, we are a strong tech community of 5000+ DS enthusiasts and professionals, conducted 20+ workshops, meetups across various cities in India and counting, 40+ contributors and proud to share that we helped 10+ folks in their Data science career.
!(https://www.datocms-assets.com/30881/1608731803-rasaworkshop-bangalore-sept-2019-grouppicture-1.jpeg ""How to build an AI chatbot using Rasa and the Zomato API" in Bangalore, India.")
Could you tell us about the challenges you have come up against during your projects and how did you manage to overcome them?
The first bot I ever created (which still supports thousands of users each day) was built using rules. Now you can imagine the performance of the bot and how difficult it can be for a developer to maintain or improve this rule-based conversation bot. However, that gave me the opportunity to prove and build an ML based pipeline for the same chatbot.
Using that raw data, I could clean, label (using NLP techniques), pre-process, solve class imbalances, and train with an ML classifier.
!(https://www.datocms-assets.com/30881/1608731808-screenshot-2020-04-01-at-13-46-10-1.png ""Machine learning pipeline to auto-tag(semi) the text data" referenced from his article on data labelling.")
Yogesh wrote about his experiments and experience of this on his Medium:
- How I used Machine learning to do the most boring data tagging job
- How I handled imbalanced text data
During those experiences, did you receive any memorable advice you'd be kind enough to share with our community?
Something I noticed along the way, building a great chatbot is not completely on the shoulders of developers. I believe the conversational design aspect is equally important and in order to do that you have to combine logic and real emotions to emulate interaction with another human.
This has drastically improved since the beginning of chatbots and therefore, I feel is changing the overall perception of them too.
Could you shed light on some of the biggest trends you see in conversational AI?
The biggest trend I have seen in conversational AI is that 'Gartner Predicts' more than 85% of businesses are adopting chatbots to automate their processes in customer support, lead generation, internal workflow etc. A chatbot's open rate via popular messaging platforms can be much higher than email or SMS. Not only that, conversational AI is changing the way we use IVR, and the introduction of voice assistants like Alexa and Google Home to every home, automobile, smart electronics, etc. has completely changed the way we interact with machines. Things we only saw in Sci-fi films of our childhoods are becoming a new reality!
!(https://www.datocms-assets.com/30881/1608731813-pyconf-hyd-dec-2019grouppicture-1-1.jpeg "Yogesh gave a talk at the 2019 PyConf Hyderabad "AI-Based chatbot building using open source bot-building framework"")
With all that in mind, how do you think technologies like Rasa can change the world?
I always believe "power is with people." In the tech world, open-sourcing a product adds great value through community contributions.
Rasa is also very well-supported by a passionate team. I believe what the research Rasa team is doing to solve the most difficult problems in conversational AI is a big factor in changing the world. While many platforms are only performing well in NLU and low code/no code platforms, Rasa is solving the actual issue that is dialogue management.
Looking to the future, could you share some of your personal aspirations?
I am learning Latin dance (Salsa and Bachata) right now! I am struggling and it feels like rocket science because I don't have any experience with Latin music and dance. Though I am pretty good at Bollywood and Indian folk (Gujarat) dance :D. But it's fun learning something which we are not good at :) It keeps me out of my comfort zone.
Soon I will be delivering more workshops online hosted by UT Dallas, you can check them out here!
We wish Yogesh the very best of luck on achieving his goals for this year, and we can't for his upcoming workshops on Rasa Open Source! Thank you, Yogesh, for all of your support!
Want to nominate someone for a spotlight? Send us an email firstname.lastname@example.org.