Skip to content

May 22nd, 2024

The Humans Behind the AI: Daksh Varshneya

  • portrait of Rasa


At Rasa, our team is building the standard infrastructure for conversational AI. Behind the scenes, the people of Rasa come together from diverse backgrounds to solve today’s most interesting challenges in conversational AI. We’re pulling back the curtain to highlight a few of the humans behind the AI.

1. Hi, Daksh. Tell us about yourself! What was your path to joining Rasa?

I graduated with a Masters’ Degree in Computer Science in 2017. With an interest in the field of Human Computer Interaction (HCI), I joined a startup in India as the first employee of the company working on diverse problems from Natural Language Processing (NLP) and Computer Vision. I was exposed to enterprise-grade problems where AI could have a real impact. One such problem was that of building reliable conversational AI to augment customer service agent’s capabilities. During the research phase I stumbled upon the Rasa framework and it immediately piqued my interest because of the flexibility it offered.

I wanted to go deeper into the NLP research area but not drop the benefits of working at a startup. This fortunately matched well with a Research role open at Rasa, which I took in 2019.

Over the course of the next 3 years I contributed to many research projects such as DIET which are now used by all Rasa users. Along the way I realized that in any machine learning company, there exists a big gap in the language business stakeholders speak and what the machine learning professionals speak. Ideally, it's on the product team to ensure that the innovation from machine learning folks is shaped correctly according to the business needs but also that the business folks understand the innovation deeply. Without an efficient bridge, there is loss in information exchange, empathy on either side and most importantly crooked alignment between teams. This is what excited me the most, so in 2023, I transitioned into a Product Management position at Rasa.

2. Take us through a typical day as a Senior Product Manager. What types of projects do you work on?

I work with the team responsible for improving and maintaining our latest innovation, CALM (Conversational AI with Language Models).

Since CALM a recent innovation, the biggest focus is on enabling a hockey stick growth curve in the adoption of CALM - not just for our existing customers, but also new prospects. I do this by interacting daily with our Customer Success team who witness our user’s reaction in the field. Apart from that, I work with other Product Managers to discover new opportunities at improving CALM further.

One thing that I cannot neglect is being up to date with the latest developments in the LLM space. Almost every day a new model gets released that beats older models and naturally becomes a candidate for us to keep our eyes on. We ensure our users are able to access it through Rasa to unlock more value out of CALM.

3. What areas of your work are you most passionate about?

Making simple solutions work for complex problems! Pure technical solutions can be quite complex at first glance. If you manage to hide the complexity neatly from the end user to provide a simple user experience (UX), that unlocks a whole lot of user value. But it’s easier said than done. This requires constantly talking to our users and understanding what works for them and what doesn’t. Sometimes it’s astonishing to see one of your initial ideas completely bashed to ground when talking to a customer but it is also rewarding since you learn a new insight about that customer’s usage.

4. What's an important problem you're solving at Rasa?

My biggest focus point right now is to enable the usage of cheaper LLMs inside CALM to bring down the overall running cost of a CALM assistant. We are solving this by exploring more efficient prompting schemes for LLMs. In parallel, we're productizing a much smaller, fine-tuned LLM that customers can reliably and securely deploy in their private environment.

Further, I am leading the efforts on enabling a scalable dialogue understanding architecture for multi-modal channels, like voice, where LLMs can be efficiently leveraged to enhance dialogue quality, but not at the cost of real time responsiveness of a conversational assistant.

5. How would you describe Rasa in three words?

Honest, Innovative, Bold

6. How do you collaborate with other teams at Rasa?

Typically, in my role, you are acting like a bridge between an engineering team and the rest of the company. The latter includes teams like customer facing engineering, sales, go-to-market, marketing, etc. I love leading cross-functional projects where the research team would quickly ship an early version of a new capability and have the customer facing teams kick the tires on it, then ensure the research team gets appropriate feedback to iterate further. This oftentimes extends to strategizing on a long-term direction with other product teams and then explaining the strategy to business stakeholders so that everyone is aligned.

7. What does a culture of diversity mean for you at Rasa?

One of the biggest motivators for me to join Rasa back in 2019 was to work in an environment that embraces people with diverse backgrounds, languages, cultures and skill sets. At Rasa, we emphasize on the value of #oneteam to succeed at our mission, but at the same time leveraging each of our diverse experiences and skill sets to execute on our mission.

8. How has working at Rasa helped your professional development?

Although I come from an ML research background, I have always been excited and curious about the problems solved by different teams that build a startup - sales, growth, product, etc. Rasa is an exceptional working place if you want to get exposure to such teams along with your primary job. The whole team embraces the culture of #oneteam and #beempowered, motivating you to not just contribute to the company's overall success, but sharpen up your skill sets in diverse domains. However, when you try to sharpen a lot of skill sets at once you often fail because you are doing too many things at once. Over time I’ve learned, and am still learning, how to prioritize the right things for the right time. That’s such a powerful skill to have in your armor for any role you play in life.

9. What's the most interesting thing you've learned lately?

Using ‘Yes, and’ v/s ‘Yes, but’ in a conversation makes so much of a difference. When someone is explaining their idea, do you add to their idea, negate it, or change the subject? It's crucial to select the right phrase to let ideas flow smoothly and, more importantly, collaboratively.

10. What's the best career advice you've received?

Building a solution to 95% completion is the easy part, the last 5% is what stretches you to your limits and is often crucial for success.

Thanks, Daksh! You can find Daksh on LinkedIn and X.

***Want to team up with Rasa? We’re hiring! Find open positions on our Jobs Board.