You’ve been recognised as Woman Role Model in ER&D. How does it feel?

I feel very honoured and humbled for being recognized as a Woman Role Model in ER&D. This award is also an endorsement of the best-in-class services QuEST Global provides to global clients from India as a team. Hopefully, this will act as an inspiration for other women to join the sector and push the boundaries of innovation.

Tell me about your role at QuEST Global and what it entails

I lead the AI Centre of Excellence (CoE) at QuEST and my primary responsibility is to provide the best solutions in Artificial Intelligence to customers and end-users. The role also calls for making QuEST an AI-aware organisation internally by setting up the Centre of Competencies to build capability and capacity in this area while extracting the best opportunities through partnerships. I am also mentoring and building a continuously evolving team of Technical Architects within QuEST to develop and deliver AI solutions in a range of technology areas, from Predictive Analytics (PA) to ML/DL to Natural Language Processing (NLP). QuEST has many industrial customers and the company has a mechanical engineering DNA. Currently, I am trying to marry industrial and AI technologies together to bring value to industrial customers.

You have had a lot of experience working in medical image processing and CV. Can you highlight the prevalence of biases in these areas of AI and efforts being made to limit these biases?

The primary advantage of AI-based medical imaging solutions in healthcare is that with these solutions we can improve and globalize healthcare while democratizing AI expertise and bringing it to even remote areas, where access to quality healthcare is poor. However, the trustworthiness and effectiveness of any machine learning solution are hugely dependent on the data that it is trained with. There is also the potential risk for biases and discrimination. The developers of AI solutions should be aware of the risks and should try to minimize potential biases at every stage in the process of product development. The quality, diversity, and quantity of data used for training the algorithms is a very important factor to be considered in order to address these risks. There should be adequate representation of the populations/subpopulations in the training dataset or else the final AI solution may false recommendations for subpopulations for which the training data was under-inclusive. Another important factor is the explainability of the AI algorithms used. Regulatory compliances and infrastructure support should be extended to ensure that the benefits of the technology should improve the lives of the people living not only in high-income countries but also in low and middle-income countries.

Can you highlight the work being done by you and your team?

As a team, we try to understand the requirements of our customers and come up with the solution, approach, and technology choices. This is achieved through consultancy services, proof of concepts, pilots, onsite AI workshops, and proactive proposals by understanding the existing business and by identifying the pain points and then providing quality solutions in Natural Language Processing, Machine Learning, and Deep Learning. We are also geared towards the implementation of best practices in these technology areas through the development of accelerators, blueprint solutions and technology demonstrators using the latest trends in AI as well as in combination with other technologies such as Augmented Reality.

What challenges do women face in building a career in emerging technologies like AI? How can the business community and society address this issue collectively and efficiently?

Managing work-life balance is a challenge that all women face. Since the technology is evolving at a very fast pace, there could be problems when woman engineers take a career break due to family commitments, maternity leave, and then come back. In such a situation, technology focused training can be offered to them so that they can efficiently contribute to the projects while continuing to build their career. Assurance and motivation from family, women-friendly policies giving importance to family and children, opportunities for higher studies can all boost the confidence in women, help them to perform better, and build a successful career in emerging technology areas.

Lack of exposure and the right mentors in emerging technology areas could be a challenge as well. But this can be solved easily in this era of digital transformation. We get opportunities to learn these technologies online, connect with masterminds in these areas and also experiment with lots of problem-solving methodologies especially in the area of AI.

What qualities do women inherently bring to the table that make them assets in a tech company?

Women are effective communicators, have an open attitude, and adapt well to changing environments. They have the skills to build an efficient team on the strong pillars of collaboration, discipline, and multitasking. They are also good listeners, innovative, and have the ability to identify the strong points of team members and blend them in such a way as to give the maximum performance and achieve success. 

Do you some positive changes / increase in female participation & representation in the tech workforce as a result of COVID?

Since the pandemic has forced the tech workforce to do their jobs from home, women get more time to be with family. She can plan activities and manage time in a more efficient manner considering the needs of family and children which makes her more productive.

What would you like to see change/improve/get introduced in the field of AI for women?

The introduction to AI should start from schools to help students get early exposure and make them understand the benefits as well as limitations of this fascinating technology. As I said before, the availability of the right mentors is an important factor. I am sure women engineers can contribute many innovative ideas. They should be given an opportunity to participate in all stages – from research to production level deployments of AI solutions in all domains so that they can come up with strategies for using AI in the future betterment of mankind.

What is your biggest AI nightmare?

AI is transforming almost all walks of life and it enables people to rethink how we can extract useful insights from data in different forms and improve decision making. But the performance of these systems depends on the data on which these solutions have been trained. The outcomes related to inaccurate data, wrong labelling, non-generalisability of the model, etc. can be serious. We must ensure that the data used for developing AI-based solutions must be of good quality, represent all possible variations, and should be adequate in quantity. For example, in healthcare algorithms should be designed with the global community in mind, and clinical validation should be performed using a representative population of the intended deployment population. This is true for any domain where AI solutions are going to be used – be it automotive, industrial or social good. Otherwise, the consequences can be very serious and can affect society in a harmful way. This is my worst nightmare.