September 25, 2024
4
min read

Coffee with Developers - Maria Apazoglou - Making AI understandable for all in production

Chris Heilmann

Hello and welcome to another edition of Coffee with Developers. Today, we're excited to share an intriguing conversation with Maria Apazoglou, a leading figure in the AI space at Thomson Reuters. Maria's career journey, insights on AI, and the exciting developments she oversees are sure to captivate both tech enthusiasts and professionals alike.

Maria Apazoglou - Making GenAI understandable for all in production

Making AI understandable for all in production (35m)

From a Young Coder to a Leading AI Expert

Maria's passion for technology began at an early age. As she shared with us, her journey into AI started quite uniquely when she was just twelve years old.

"The first thing that I did when I was twelve years old is a regression model for predicting sales."

Growing up with a love for coding, Maria naturally found herself gravitating towards data and analytics. Her career has been a tapestry of developing models for traders, overseeing data platforms, and AI solutions across various sectors including energy and financial services.

Career Milestones

Maria's career trajectory is filled with notable positions and responsibilities:

  • HSBC: Before joining Thomson Reuters, Maria served as the Chief Data Officer at HSBC, where she led various data and analytics platforms.
  • Thomson Reuters: At Thomson Reuters, she spearheads the AI platform, data platform, and oversees CoCounsel, one of their flagship AI products.

Maria's work has always been about harnessing data to derive actionable insights and creating platforms that democratise AI usage.

Thomson Reuters and the AI Revolution

Thomson Reuters has been a significant player in the AI and data analytics space, and Maria's work is central to their mission of leveraging AI for broader accessibility and efficiency.

Building an Inclusive AI Platform

Maria emphasised their focus on not just catering to data scientists but also enabling non-technical users to create and utilise AI solutions effectively.

Key Principles of the AI Platform:

  1. Self-Service Orientation: The platform is designed to be user-friendly, allowing users to interact with large language models and leverage them for various tasks through an intuitive interface.
  2. Security-First Approach: Users’ data security is paramount. The platform ensures that even the centralised platform team does not have access to the user's data.

Overcoming Challenges in AI

Maria discussed some of the prevalent issues in AI such as data quality, user fear and confusion, and the notorious issue of AI "hallucinations" – where models generate incorrect or nonsensical information.

Strategies to Tackle Challenges:

  1. Educating Users: Through extensive training sessions, user stories, demos, and hackathons, the team at Thomson Reuters educates users on AI capabilities and best practices.
  2. Transparency and Security: By making it clear that neither Thomson Reuters nor the platform team access users' data, they build trust and encourage broader adoption.
  3. Promoting Critical Thinking: Maria believes in fostering a critical mindset when interacting with AI, ensuring users cross-check information and understand AI’s limitations.

User Success Stories

The fruit of these efforts is evident. Users across the organisation have been leveraging AI for a myriad of purposes:

  • Sales Teams: Utilising AI to draft responses and summarise documents.
  • QA Engineers: Automating tests and improving efficiency.
  • Personal Productivity: Maria herself uses AI to write appreciation emails, a task she admits isn't her strong suit.

These success stories inspire others and show tangible improvements in efficiency and creativity across the board.

Navigating the Complexities of AI Usage

The AI landscape is dynamic, and keeping up with the latest developments is crucial. Maria shared the evolving nature of their platform and the strategic decisions they've taken to ensure scalability and robustness.

Handling AI’s Limitations

Maria demos the AI capabilities, deliberately showcasing its limits to set realistic expectations.

Example Demo:

  • Asking an AI model about the current date to demonstrate that the model’s knowledge is bounded by its training data timelines.

Encouraging Efficient Utilisation

To drive efficiency, the platform allows users to create and save customised AI solutions, reducing redundant prompts and saving valuable time.

Example Workflow:

  • Users can store a series of instructions or prompt chains that they frequently use, making future interactions with AI faster and more streamlined.

The Role of Diverse Teams in AI Development

Diversity in teams is not just about gender but spans across backgrounds, ethnicities, and thought processes. Maria’s teams at Thomson Reuters exemplify this diversity, bringing various perspectives that improve the overall quality and user-friendliness of AI solutions.

"The diversity of thinking is one of the things that I value the most."

The Future of AI in the Workforce

Despite the buzz around AI, Maria is pragmatic about its current state and potential. She sees AI as a tool that augments human skills, allowing more people to enter the tech field from varied backgrounds and disciplines.

AI as an Enabler

AI tools can demystify programming and data science, making them accessible to non-experts. This has profound implications for workforce diversity and the democratisation of technology.

Examples of AI in Action:

  • Legal Backgrounds: Professionals with a legal background transitioning into technical roles, using their domain expertise to handle AI and data tasks effectively.
  • Prompt Engineering: Crafting specific AI prompts that leverage subject matter expertise to get more accurate and useful outputs.

Sharing Knowledge and Best Practices

Maria and her team are committed to sharing their insights and learnings with the community. They actively participate in conferences like re:Invent and publish blog posts detailing their architectural and strategic decisions.

Upcoming Initiatives:

  • A detailed blog post on how Thomson Reuters democratised AI capabilities.
  • Continuous contributions to the community through presentations and collaborative discussions.

Conclusion

Our coffee with Maria Apazoglou was as enlightening as it was engaging. From her early forays into coding to her current leadership at Thomson Reuters, Maria’s journey is an inspiring tale of passion and innovation in the AI space.

If you want to dive deeper into Thomson Reuters’ AI initiatives or explore career opportunities within their dynamic teams, visit their AI section on the Thomson Reuters website.

Maria's work is a testament to the fact that with the right tools and mindset, the future of AI holds immense potential for innovation across diverse sectors. Stay tuned for more Coffee with Developers sessions as we continue to bring you insights from the brightest minds in technology.

Coffee with Developers - Maria Apazoglou - Making AI understandable for all in production

September 25, 2024
4
min read

Continue reading

We are busy writing more posts on this topic right now. Sign up for our newsletter to not miss them.

Subscribe to DevDigest

Get a weekly, curated and easy to digest email with everything that matters in the developer world.

Learn more

From developers. For developers.