The People Behind The Bots — Diana Mingels
Diana Mingels is the Senior Manager of Conversational AI at Capital One. Currently, Diana leads the engineering teams working on AI capabilities for Eno, Capital One’s Intelligent Assistant. Diana has been working in the field of NLP for an impressive 14 years and has worked for companies all over the world including Nuance Communications. We hear all about Diana’s journey into the world of language and Conversational AI, let’s get straight into it!
Q1 — What drew you to the Conversational AI industry?
Diana: Since childhood I have had a passion for mathematics and languages. That resulted in me joining a Mathematical Linguistics department for my undergraduate studies, followed by 2 Masters Degrees in Mathematical Linguistics and Natural Language Processing/Human language Technology. As a result, the Conversational AI industry was a natural continuation of my career given my education, background and interests. I guess I got lucky that such a perfectly matching industry emerged just about when I was graduating from my studies. Ever since, I have been loyal to the Conversational AI space and have contributed to most used Intelligent Assistants known across the globe.
Q2 — What skills or transferable skills helped you when entering this space?
Diana: I think for any AI project, and maybe even more so for Conversational AI, the strongest asset you might possess is an interdisciplinary background. Having lived and worked in 6 different countries over the past decade and having routinely traveled to countries in Europe in the US, I gained a lot of experience and knowledge of communicating and interacting with different cultures in different languages. When building Conversational AI, I strive to translate my knowledge into software and teach Intelligent Assistants to do the same. My linguistics background also allows me to see a clear distinction between where it’s best to leverage deep learning vs where traditional rule-based approaches are still unbeatable.
Q3 — Were there any specific resources, tools, industry experts, that helped you along the way?
Diana: I learned a lot “on the job” from my colleagues at Prompt, Nuance, Axafina and Capital One. I have also dedicated a lot of time to continuously researching and experimenting in my free time. I have completed various specializations ranging from Java programming to Deep Learning and NLP. Every single one gave me some new knowledge, confidence and skills. Also, being subscribed to mailing lists in your fields of interest is the best way to stay in touch with quickly evolving research and industry trends, and that’s something that helps me to be up to date on events and developments happening in the Conversational AI space. I try to regularly attend conferences and talk directly to the authors of the work that is of most interest and relevance to my projects. You can’t be an expert without knowing what’s going on outside of your project and daily work.
Q4 — In a sentence or two tell us what your role entails
Diana: I lead engineering teams working on AI capabilities for Capital One’s Conversation Platform, powering our Intelligent Assistant Eno. My responsibilities encompass various types of work, including research and industry benchmarking, strategic planning, innovation, assessment of technology, establishing and improving software and ML life-cycle processes, feature work and run-the-engine activities, as well as sharing the knowledge with the stakeholders of various backgrounds, and people leadership.
Q5 — What do you enjoy most about your role?
Diana: I enjoy working with talented and motivated people willing to innovate, create value for our customers and build the future of Conversational AI. I enjoy learning about new advancements and trends in tech, products, emerging start-ups, assessing and evaluating industry trends, pushing the boundaries and bringing awareness on how quickly modern technology is evolving and how to ensure our success in that highly competitive environment.
Q6 — What is the most challenging part of your role?
Diana: Working on innovation in a big company is always a challenge. Not only do you need to be skilled at understanding the technology and what’s possible, but also being able to convince less tech-savvy audiences and decision makers. There is always a trade-off between long-term investment and short-term benefit. Finding the right balance to make sure innovation doesn’t die out and measurable progress is being made is difficult.
Q7 — What excites you about the future of this industry?
Diana: This industry is evolving rapidly. Conversational interfaces are the future of how customers are going to interact with businesses. Hence, they are the future on how people will interact. Expectations of customers are growing at a high pace fuelled by better offerings and improved accuracy. I think in 5 years from now the customer-business interaction model will change drastically thanks to the advancements in ASR, NLU, NLG, TTS as well as new approaches to dialogue management and enablement of better conversation memory. Many people might find it hard to believe, but I am betting that we are going back to doing everything with our voice, texting will disappear into the thin air sooner than we think.
In the nearest future we won’t need to interact with rigid and limited IVR systems that don’t understand you or tell you things you don’t want to hear. Flawless customer-centric experiences empowered by advanced Conversational AI techniques are going to help people to release their precious time on more productive and positive activities than spending hours on customer support. That applies to both customers and businesses, a win-win situation for everyone.
Q8 — What is one piece of advice you would give to other people looking to enter this industry?
Diana: A lot of research on modern Conversational AI is formulated around generative dialogue and a single end-to-end deep learning model that can converse with the users on a wide range of topics and keep them engaged.
Deep learning allows to unlock a lot of capabilities and scale in NLP tasks, but it can’t solve complex integrations with DBs, business logic and responses that require high controllability and predictability.
While there is some value in open-ended dialogue, the goal-driven or task-oriented systems are still the most valuable applications of Conversational AI to solve business problems.
Hence, my advice is to be open-minded and learn to use the right technology for the right task. Real-world Conversation systems rely on a combination of different techniques and technologies and to be able to build an end-to-end system you need to take time to understand the complexity of all pieces and parts involved in the pipeline, rather than blindly believing in a magic “end-to-end” solution that creates itself based on dialog data alone.
Techires’ takeaways:
- Having an interdisciplinary background is well suited to the Conversational AI industry!
- Researching and experimenting in your spare time pays off.
- Don’t limit yourself to a certain technology, use what’s best for the job!
Thank you for reading! A special thanks to our guest, Diana. We really enjoyed hearing Diana’s insights, and hope you did too! Click here to follow Diana on LinkedIn!
If you’re keen to hear more insights and advice from industry experts in the Conversational AI space, don’t forget to click the follow button to be notified of future posts!
You can also head over to our socials for more tips and advice from the team at techire ai and if you’re looking for roles within Conversational AI check out our jobs page and website.
And if you’d be interested in taking part in our blog series get in touch, we’d love to hear from you!
Allys Parsons — Conversational AI Recruiter @ Co-Founder @ techire ai
Leigh Smith — Conversational AI Recruiter @ techire ai
