November 25, 2024 Mark Ginnebaugh

Connie Yang on "Women Leading the Charge in Open Source AI"

Connie Yang, DesignMind VP Data Science and ML

Connie Yang, VP of Data Science and ML at DesignMind, was asked to share her thoughts about women in the Open Source AI space for emPOWERED magazine. The interview is published here with permission.

Tell us about your journey in open source AI

I started my career as a data scientist at Microsoft. I did my undergrad in math and computer science at Carnegie Mellon and did a data science internship at Microsoft before going full-time. Open-source technology, especially AI, has been a critical part of my work. During my time at Microsoft, there were early developments of InterpretML, the open source technology for explainable AI and so there are a lot of very exciting things in that space that we were able to leverage to help explain our work.

Currently at DesignMind, we're using a lot of open source AI models like Llama and Mistral to develop generative AI retrieval augmented generative (RAG) solutions for our large enterprise customers.

Throughout my career, I've leveraged a variety of open source tools to build scalable AI solutions, from machine learning frameworks to cloud infrastructure platforms. Currently at DesignMind, we're using a lot of open source AI models like Llama and Mistral to develop generative AI retrieval augmented generative (RAG) solutions for our large enterprise customers.

I've been focused on evaluating a lot of open source AI to assess the accuracy of generative AI applications. I'm frequently active in GitHub repositories, and it's an incredible resource for collaboration where I'm able to engage with like-minded experts from around the world, whether we're troubleshooting issues or exploring interesting findings in the application of these models.

Why do you think there’s a lack of women in open source development and open source AI? What are the main barriers preventing women from participating in open source AI

One significant issue is visibility — there just aren't enough high-profile female role models in this space, which makes it difficult for women to see a clear path for themselves. I also think the culture of open source development can feel like a boys club, where a lot of the competitiveness can overshadow collaboration, and that makes this space feel like less welcoming.

Timing is another big challenge too. Open source work is done on a volunteer basis, outside of your normal working hours, and this can be particularly difficult for women who are balancing other responsibilities, for example, for women who become new mothers, that available time shrinks even further.

I co-founded Microsoft's Women in Data Science Community. We started the community because we were often the only females. In fact, I was the only woman on my team reporting to my manager with 13 other data scientists who were men.

I co-founded Microsoft's Women in Data Science community. We co-founded the community because we were often the only females. In fact, I was the only woman on my team reporting to my manager with 13 other data scientists who were men.

We built the grassroots movement up from just the two of us to now a 1,000-plus community. My fellow co-founder is still at Microsoft and we still hold panelist talks and workshops at the internal machine learning and data science conference every year.

We've heard from women who want to contribute to the open source space in machine learning and AI and have talked about how difficult it is to juggle the work responsibilities if they were to go on maternity leave and how that affects their careers. A lot of the cultural and structural factors often discourage participation and can limit a lot of opportunities for women to gain visibility in the community.

What can be done to improve and encourage more women in AI, data science, and wider open source development?

My big thing is hackathons — they are a great way for people to foster collaboration and to come together and work in real-time.

I've been a part of enough hackathons to know that most great ideas stemmed from some hackathon. To hold more hackathons to encourage women to participate in major projects like responsible AI, as a big part is people don't know which projects they can contribute to.

Hosting a hackathon for such projects would encourage people to take part, and then in touch through Discord channels.

Finally, who is your female open source hero?

My female open source hero is Margaret Mitchell, who used to be the CO lead of Google's ethical AI team.

My female open source hero is Margaret Mitchell, who used to be the CO lead of Google's ethical AI team. She has been a real driving force in advancing a lot of AI ethics. Her work on open source projects related to explainable AI has been instrumental in making like a lot of the AI systems we see today more transparent and accountable. Her commitment to building tools that prioritise fairness and transparency aligns perfectly, with the open source philosophy of collaboration and shared progress.

Create a Custom Copilot Using Microsoft Creator Tools Another person who I have a lot of respect for is Dr. Rachel Thomas, who co-founded fast.ai. She has done a lot of work in democratizing AI education and accessibility within the space. She's significantly lowered the barrier of entry for machine learning, making it a lot easier for women and underrepresented groups to engage in AI and I think that somebody like her should follow suit doing that for the open-source community in AI and lower the barrier of entry for people of diverse backgrounds to be able to contribute to these projects.

Connie Yang is VP, Data Science and ML at DesignMind.

Please check out our AI and Data Science Solutions, including Machine Learning, LLMs, and AI Strategy. You can view additional AI and Data Science articles by our experts here.