Women Leading The AI Industry: Why and How We Can Solve The ‘Leaky Pipeline’ Problem in Tech, With Kalai Ramea of PARC

Tyler Gallagher
May 19 · 10 min read

It is well-known that there is a diversity problem in AI (and technology fields). Contrary to popular opinion, not having enough women in technology is not a pipeline problem. There are many qualified women — it is, in fact, the “leaky pipeline” problem, where women are leaving in droves due to their bitter experience in workplaces. To prevent that, companies should actively train their employees to recognize their unconscious biases, systemic biases (class, race, gender, etc.), and microaggressions.

Moreover, diversity (in numbers) alone is not sufficient; there needs to be equal emphasis on inclusion. There needs to be an active introspection to understand if the minority members feel included in teams and activities.

Companies should recognize that they benefit from having a diverse group of experiences and ideas to build successful products. So, it is of their advantage to hire more women, people of color, LGBT folks, and other minorities.


As part of my series about the women leading the Artificial Intelligence industry, I had the pleasure of interviewing Kalai Ramea, Data Scientist at PARC, a Xerox Company.

Kalai Ramea is a data scientist at PARC, who focuses on statistical machine learning and data analytics in various domains. Her research interests include applied machine learning and deep learning for scientific domains, exploration of novel statistical modeling techniques, and big data analytics. Prior to PARC, Kalai worked as a researcher at the University of California, Davis, where she developed several numerical models in the domains of energy, transportation, and climate, in order to assess long-term policy impacts of energy-efficient technologies and human behavior. Kalai also worked as a research fellow at the International Institute for Applied Systems Analysis in Austria as part of the Young Scientists Summer Program, where she developed an energy systems model that incorporated consumer behavior to analyze long-term climate impacts. She also has 3 years of consulting experience in engineering design. When not working with data, she draws comics and volunteers at pet rescue centers.


I have always been passionate about mathematics. Even as a child, that was the one subject I did not complain about studying. I did my Bachelors and Masters in Engineering and worked as a consultant for a few years. But, I was not satisfied with that job, as my mind was longing to solve math-oriented problems. So I went back to school to do a Ph.D. where I developed quantitative models for energy and climate systems. When I finished my doctorate, I discovered the wonderful world of data science and AI and felt that the synergy between problem-solving, math, and coding is the perfect place to pursue my career. And I have never had a dull day since then!

Follow your passion! It may sound simple, but that is crucial to be happy with your career. Not everyone needs to be good at math, or needs to code, or needs to pursue science. But, if you do like those things, now is a great time to get into data science as we have so many exciting problems to solve. Also, it is okay to take the ‘scenic route.’ I did not do a Ph.D. immediately after my Masters. I dipped my toes in consultancy, learned so much in that position that I still use to this day. My Ph.D. in applying mathematical models to climate and energy systems gave me a window to know about climate urgency, and I hold it close to my heart. So, it is never a ‘waste of time’ to gain an experience that is unrelated to where you are currently.

One of the best things about working at PARC is the ability to work on a variety of projects. Last year, I developed an AI algorithm to generate quick and unique artwork. This year, I am working on a project to develop a pipeline to measure atmospheric gases from satellite and ground-based observations. Monitoring atmospheric gases is a huge challenge in general, and specifically, we need better accounting for greenhouse gases (such as methane and carbon dioxide) to regulate the emitting industries. AI algorithms can help in processing the massive amounts of data coming from satellites, to perform bias correction, and to upscale the resolution. As I explore this domain, I am excited to see how we can improve the traditional approaches using AI.

When I was in high school, my Computer Science teacher Mrs. Mallika Ranjan discovered that I had a passion for math and coding and encouraged me to push further in that direction. She used to say, “with your creativity, you can do wonders in computer science.” At that time, my parents were not financially equipped to get me a personal computer, so she kept the programming lab open for me after school hours so that I could practice coding and do side-projects! I am incredibly grateful to her for that. And I believe this experience was a huge catalyst for me to develop a positive attitude towards coding (and computer science in general), which led me to where I am today.

  1. There are novel ways of data collection now (sensors and Internet of Things, massive amounts of imagery/text, multi-modal data fusion, etc.), and AI algorithms, along with improved computing power, can ingest and process them, which could not have been possible a few years ago.
  2. Scientific equations in traditional physics models can now be powered by deep learning for faster computation. This has significant implications in domains such as climate science and particle physics, where the models can take days to weeks to run simulations, which can now be solved in hours with the help of AI. We are starting to see an increasing amount of applications in this area (Machine learning for Science).
  3. Natural language processing had a great year with an exploding amount of research advancements in pre-trained language models. This has the potential to revolutionize language understanding the same way deep learning advancements have been revolutionizing computer vision.
  4. We also see an increasing amount of democratized learning platforms in the form of MOOCs such as fast.ai, Coursera, and edx, which allows anyone to learn the fundamentals to get into this domain. This sort of inclusivity is necessary to further the field in the right direction.
  5. AI, in general, has been recognized as one of the top research areas by government and private entities, and we see several significant funding sources to support researchers.

  1. Anyone working in the AI domain can tell that there is a lot of hype around machine learning and deep learning algorithms. To their credit, these algorithms have made tremendous advancements, but they still have many limitations. Many AI practitioners tend to think that acquiring massive amounts of data and applying deep learning algorithms would solve the problem. But, real-world data needs a lot of creativity in problem-solving and collaboration with domain experts before we get to the algorithm development stage.
  2. As engineers and scientists, we get too carried away by the technical approaches and forget to focus on the ethics of the solution. Some examples of problematic solutions are facial recognition applications, predicting prison recidivism, and synthetic face generation (“deep fakes”). These applications impact the most vulnerable groups when deployed. This is why it is vital to have ethics researchers on board while developing and implementing an AI application for the real world.
  3. AI researchers and practitioners need to understand that there is an enormous digital divide in the world. Most of the global south do not have access to smart devices where such AI algorithms would be deployed. Being in Silicon Valley, we are conditioned to come up with fancy solutions to real-world problems, but in practice, low-technology solutions may suit better to the particular problem. So, we need to be aware of the ground reality when thinking about solving a problem.
  4. There is a lot of ‘AI for social good’ applications that are being developed in recent times. While this is a welcome step, AI practitioners should consider the implications of imposing a technological solution to the vulnerable group. For instance, a GPS tracking project of endangered animals can go haywire when poachers hack the system.
  5. Finally, as AI is penetrating different domains such as healthcare, environment, and social sciences, it is more important than ever to listen to the domain experts and center them while developing any AI application for these domains.

I do not believe AGI (artificial general intelligence), where a robot achieves human-level intelligence to destroy the human race, is a possibility (at least not right now or in the near future). We certainly have some legitimate concerns with AI, in terms of ethics, privacy, and fairness, as I described above. Those are the more pressing concerns to humanity with regards to AI.

Scientists and engineers who excel in algorithm development cannot look through the lens of anthropologists or social science scholars. We need to bring in a diversity of ideas as well as experiences to add that dimension. That is the only way to alleviate these bias and fairness concerns and make our AI systems more robust.

More than my ‘individual’ success story, I would like to tell a success story about an all-women team that came together for a hackathon to build a product to help people unemployed due to COVID-19. A group of us (eight women with different backgrounds — AI, social science, natural science, engineering, business, and design) participated in ‘Women Hack the Crisis’ event organized by The Expat Woman. This was a week-long event, and the idea was to develop a solution for people affected by COVID-19.

We have been reading that an unprecedented amount of people are losing their jobs, and we discovered that millions of people are not getting the help they need, as the system is extremely complicated to navigate. So our team decided to build a conversational AI chatbot that would help people answer important questions on unemployment benefits and direct them to the right resources.

In just four days, we read through several stories on social media to identify the pain points of people filing unemployment and built a knowledge base of user-profiles with flowcharts for about 50 questions. We then developed an interactive chatbot called BEBO (a shortened version for Benefits Bot) in two languages: English and Spanish. You can find our website here where BEBO can answer questions on benefits: https://whowteam.github.io/bebo/ A demo video of BEBO can be found here: https://youtu.be/_CKUKD1InmM

  1. Data science (and AI, in general) is a crowded space. So, do not be discouraged by failures especially when you are trying to get into the industry
  2. Actively find allies and talk to other women in this field. There are several chapters of Women in ML and Data Science groups all over the world.
  3. When you are new to the field, having a mentor who can help navigate this space helps a lot

It is well-known that there is a diversity problem in AI (and technology field). Contradictory to popular opinion, not having enough women in technology is not a pipeline problem. There are many qualified women — it is, in fact, the “leaky pipeline” problem, where women are leaving in droves due to their bitter experience in workplaces. To prevent that, companies should actively train their employees to recognize their unconscious biases, systemic biases (class, race, gender, etc.), and microaggressions.

Moreover, diversity (in numbers) alone is not sufficient; there needs to be equal emphasis on inclusion. There needs to be an active introspection to understand if the minority members feel included in teams and activities.

Companies should recognize that they benefit from having a diverse group of experiences and ideas to build successful products. So, it is of their advantage to hire more women, people of color, LGBT folks, and other minorities.

I would say this quote has resonated with me many times: “Life is not about waiting for the storm to pass. It’s about learning to dance in the rain.” As someone with a “Type A personality,” I used to (and still) get work-obsessed if something needs to be done, and would try to do it as quickly as possible. I had to actively learn to slow down and enjoy the process and do it mindfully. I believe there is a lot of learning and joy in slowing down, and especially, science, in general, needs time to think and reflect.

To bring my passion and experience together, I would start a movement to build ethical AI systems for natural and social sciences, to tackle climate change and environmental justice. And, many excellent researchers are already pushing boundaries on this idea.

They can follow me on Twitter: @KalaiRamea

Thanks for a fantastic set of questions. I really enjoyed answering them!

Authority Magazine

Leadership Lessons from Authorities in Business, Film…

Tyler Gallagher

Written by

CEO and Founder of Regal Assets

Authority Magazine

Leadership Lessons from Authorities in Business, Film, Sports and Tech. Authority Mag is devoted primarily to sharing interesting feature interviews of people who are authorities in their industry. We use interviews to draw out stories that are both empowering and actionable.

Tyler Gallagher

Written by

CEO and Founder of Regal Assets

Authority Magazine

Leadership Lessons from Authorities in Business, Film, Sports and Tech. Authority Mag is devoted primarily to sharing interesting feature interviews of people who are authorities in their industry. We use interviews to draw out stories that are both empowering and actionable.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store