Machine learning engineer Tanya Dixit ventured from India to Australia in 2019 to join one of the few graduate programs in the world focused squarely on machine learning.
She also wanted to reboot her life.
During nearly three years as an Embedded Systems Engineer in Hyderabad in southern India, she had felt uninspired and adrift until a midnight epiphany inspired her to return to the academic setting.
“I was looking at the surrounding ecosystem, not just the university,” Dixit said.
As she researched Australia’s capital city, she was enticed by a burgeoning network of startup enterprises, incubators and development hubs. At its epicentre was The Australian National University (ANU).
“I wanted to be in a place that is growing and thriving. I wanted to be around people working in cutting-edge technology and bringing new ideas to life,” Dixit said.
“I thought, if I could be around people like that, eventually I could be one of them.”
Shy girl’s refuge
Dixit looks back on her childhood and early career in India with mixed emotions. She’s proud of her new life and career in Canberra. But she wonders, had she stayed in India, could she have slayed her dragons?
“I had always been this introverted, shy girl who didn’t know how to talk to people. I never believed in myself,” Dixit said.
Perceived as awkward and aloof in group settings, she recalls asking people to talk with her about big ideas, and being criticised for being “supercilious”.
“I like to know what people really ache for, and some people just do not want to talk about that. So I learned to be quiet,” she said.
Mathematics and science were her refuge. She sought out learning materials outside her school curriculum, such as academic journals and online courses.
“I loved learning about maths and science. There was always more to explore and that fascinated me,” she said.
Although her first love was mathematics, Dixit’s father often repeated the phrase, ‘Technology is going to change the world’.
“I often wondered about how I could change the world. So, I decided to pursue a career in tech,” said Dixit.
Navigating an undergraduate degree in electrical engineering at Birla Institute of Technology and Science, Pilani, and entering the Indian workforce, she was confronted by patriarchy and a jarring gender gap.
“I felt like I was put under a microscope because I was female. I hated it,” she said. “It’s hard to explain how it feels when you can’t allow yourself a single mistake.”
Dixit describes her time as an audio codec engineer as “dark days in my life”. She struggled to stay motivated, and worried that she’d chosen the wrong career.
“When I am lost and sad, I buy books or [online] courses because they sort of push me to evolve,” Dixit said.
In 2017, she purchased an online course called Udacity’s Deep Learning Nanodegree, which has since been folded into another course on artificial intelligence (AI).
This was her introduction to the backpropagation algorithm which is widely used to train deep learning models.
“I was in my room watching a video on my laptop that explained the math behind it, mainly calculus, and I was blown away. It was beautiful,” Dixit said.
She replayed the video a few times before she understood the algorithm. But once she did, something clicked.
Backpropagation reconnected with her love of mathematics, and in the same instant, it revealed an ever expanding universe of applications for machine learning in the near and distant future.
“The more I delved into machine learning, the more I realised that I had found something to keep challenging me for the rest of my life. I also saw the potential impact of this technology. I could see myself solving problems I have always wanted to solve,” she said.
A close-knit community of nerds
Dixit arrived in Canberra in July 2019 to pursue a postgraduate at ANU.
“As soon as I got here I felt the pressure come off,” Dixit said. “I was finally on my own in a new country, I could just be myself and no one would care. I found that very liberating.”
She quickly connected with “key mentors” at the ANU School of Computing, including Professor Stephen Gould, Professor Steve Blackburn, and Dr Alex Antic. Gender was no longer an issue, and she found it easy to interact with her peers in the Master of Machine Learning and Computer Vision program.
“People think we’re all nerds who don’t like to talk to people. That may be true for some, but the collaborative aspect of coding is one of the most exciting parts for me,” she said.
When asked what she learned in her two years earning a Master of Machine Learning and Computer Vision at ANU, she talked about “soft skills” at length before getting to the “hard skills” such as software engineering.
“I learnt to interact with people from different areas, not just machine learning, not just computer science. People from all different kinds of professions, all different areas of life,” she said.
Dixit found that collaboration and communication were essential skills in her studies at ANU, and later her profession.
“Machine learning is a very math heavy field. In order to talk to someone who’s not familiar with machine learning, you need to distill those concepts and then talk to them in their language. ANU taught me how to do that.”
“This has proved really valuable in my career. That’s a highly valued skill.”
Another thing ANU taught her is courage.
“I remember I was in Professor Steve Blackburn’s class and he was teaching us about recursion and then he asked, ‘What’s the most important ingredient for a good coder?’
“We were all thinking it must be something very technical. But he said, ‘It’s courage.’ You need to have courage to be a good coder, programmer, or computer scientist. Because when you’re writing that code, you need to have courage in yourself to keep going if you are stuck, and then to believe that it’ll all work out.
“And if it doesn’t work, you always have yourself, knowing you can work through the problems.”
Another concept Dixit underscored was ‘ownership’, because data scientists are responsible for solving business problems, not just technical ones.
“You have to own the whole cycle. A lot of people think of themselves as software engineers only. But that’s not true. A data scientist is someone who understands the entire business problem, someone who interacts with the subject matter experts, and someone who owns the solution.”
As for the hard skills she gained at ANU, Dixit developed an understanding of calculus, algebra statistics, software engineering, databases, and knowing how to deploy a mode.
Her professors and tutors, she said, seemed personally invested in her professional and personal development. She was so grateful to her tutors, in fact, that she decided to become one herself as a way to give back.
“Coming to ANU and interacting with my professors, interacting with my peers, it gave me that confidence to really trust my voice. It was a life changing experience for me.”
Dream job across the lake
“When you apply machine learning to a new industry, there is always a lot of uncertainty. It’s the not-knowing that I love the most. It’s thrilling,” Dixit said.
Helping companies understand how to gain a competitive edge using machine learning is now part of Dixit’s remit as a Machine Learning Engineer. She works at the Crayon AI Center of Excellence, an international company with an office adjacent to Parliament, across the lake from ANU.
“I love the sheer challenge that it presents, because machine learning is an ever changing field and you see it being applied everywhere. The research is so fast-paced that it changes almost every day.”
The innovation ecosystem Dixit sought proved instrumental in finding the inclusive and dynamic professional environment she envisioned during her ‘dark days’.
She was introduced to her first employer in Canberra at an ANU job fair. But ANU was a bridge for Dixit in more ways than one.
No longer the shy introvert, Dixit serves as the editor and chief of Canberra’s first AI newsletter The Future Stack. She frequents networking events and often meets up with friends for coffee and conversation. She’s also started her own company.
“I can be quite an extrovert when I feel comfortable being myself,” she said.
State of the ANU, Canberra ecosystem
Professor Gould has stayed in touch with Dixit after graduation and is proud of her success. The innovation ecosystem that attracted Dixit to Canberra is growing rapidly, he said, and has tremendous potential.
“If you look at the most innovative centres of technology around the world, they are all co-located around world-class research-intensive universities feeding students and ideas into a commercial ecosystem,” he said, citing Stanford University in California’s Silicon Valley, UT Austin in Texas, Technion in Israel, and Cyber Valley in Germany.
With the right government economic incentives, Canberra could well join their ranks.
The ANU is strengthened by its proximity to CSIRO (the Commonwealth Scientific and Industrial Research Organisation) and the Canberra Innovation Network (CBRIN), while the Centre for Entrepreneurial Agri-Technology (CEAT) is housed on campus just a stone’s throw from the School of Computing.
Both Dixit and Gould mentioned CSIRO’s Data61 which, until the pandemic, had been housed on campus in the Computer Science and Information Technology Building. Gould said the longstanding close relationship between Data61 and the ANU is one of the bedrocks of Canberra’s technology and innovation nexus.