Balancing Speed and Efficiency in AI Development: Insights from the Head of Data Science at Chime
AI leaders constantly face the challenge of deciding whether to build AI solutions in-house or to purchase them from vendors. During our panel discussion at AI Dev Summit 2024 in San Francisco we gathered AI leaders from PayPal, Chime, and Gusto to uncover the decisions these leaders go through when deciding on building vs buying AI solutions. The panel was moderated by Remy Thellier, Head of Partnerships at Vectice.
This is the second article in our three-part series, in which we uncover the insights of Philip Denkabe, Head of Data Science at Chime, who shared his approach to prioritizing speed and efficiency in AI development.
You can read the first articles with Khatereh Khodavirdi, Vice President of Consumer Global Growth Enterprise for Data Science at PayPal on Navigating the AI Build vs. Buy Decision
About Philip Denkabe
Since joining Chime in December of last year, Philip has been pivotal in rapidly building and scaling the data science team. His role encompasses constructing a robust team, devising strategic focus areas, and leading innovative research to uplift Chime’s business.
Under his leadership, the team aims to develop several types of data science models spanning generative AI and non-generative AI models.
Remy Thellier’s key Takeaway
“From my conversation with Philip, I gained several key insights for teams growing fast:
- First, conducting a thorough cost-benefit analysis is essential to prioritize projects. This helps us understand resource requirements, timelines, and potential business impacts, considering both direct and indirect costs.
- Balancing short-term and long-term planning is crucial. We should identify quick wins that deliver immediate business value while also planning for future growth and aligning with the company’s broader vision.
- It’s important to evaluate the efficiency of using existing solutions versus building from scratch. We need to weigh the benefits of internal development against the convenience and speed of external platforms.
- Building team capabilities is another key point. We should demonstrate value with existing team skills before acquiring new resources. This fosters a culture of growth and continuous learning, while strategically hiring to fill skill gaps.
- Incorporating a level of abstraction in our tools can enhance speed and efficiency. By focusing on core tasks and leveraging pre-built components, we can speed up the development process and avoid wasting time and money on non-core features.
- Finally, efficient tool usage is critical. Streamlining processes by using efficient tools and external solutions for non-core functionalities allows our team to concentrate on delivering faster results.”
Philip’s Top 6 Tips for Leaders Making Build vs. Buy Decisions
1. Conducting Cost-Benefit Analysis
Q: How do you prioritize tasks and projects in your context?
Philip stresses the importance of conducting a thorough cost-benefit analysis for each AI project. This analysis helps in understanding the resources required, the time it will take, and the potential business impact. By evaluating these factors, teams can prioritize projects that deliver the most value.
“We conduct a cost-benefit analysis for everything we do. First, we identify our goals, assess our current position, and then define our target. Finally, we determine what is needed to move from point A to point B.”
He also emphasized that this process involves looking at both direct and indirect costs, including the costs of hiring new resources or building new platform features.
2. Short-Term vs. Long-Term Planning
Q: How do you approach both short-term and long-term planning?
Balancing short-term goals with long-term strategic planning is crucial. Philip discusses the need to identify quick wins that can demonstrate immediate business value while keeping an eye on long-term objectives.
“It’s both short-term planning and longer-term planning: identifying the resources that we need, determining what business goals we need to meet, and then planning accordingly.”
Philip added: “We tend to chew gum and run simultaneously, building a function that should support all of the different functions within Chime. But at the same time, we also have to strategically think about where we want to be two, three, four years from now.”
This dual approach ensures that the team remains aligned with the company’s broader vision and can adapt to changing market dynamics, effectively managing both immediate needs and future growth.
3. Leveraging Existing Solutions
Q: Can you share an example of a build vs. buy decision you’ve faced recently?
In some cases, using existing solutions can be more efficient than building from scratch. Philip shared an example of deciding between training models internally or using AWS services.
“We’re currently exploring Generative AI and deciding on our approach. One option is to build a solution internally, which involves acquiring a platform and training our models from scratch, despite the costs. Alternatively, we could use AWS for a simpler and quicker implementation.”
This decision-making process involves weighing the benefits of internal development against the convenience and speed of using external platforms.
4. Building Team Capabilities
Q: How do you handle skill gaps and ensure your team can grow with the company’s needs?
Philip highlights the importance of using existing team skills to demonstrate value before seeking additional resources. This approach helps secure decision-makers’ buy-in and justifies the need for new skill sets. It also fosters a culture of growth and continuous learning within the team.
“We aim to leverage our existing internal skill sets to demonstrate value rather than acquiring new ones. Since we are not an academic department, we must show tangible results.”
Philip also emphasizes the need for strategic hiring to fill skill gaps and enhance team capabilities: “It’s not about having all the skill sets in-house from the start. It’s about proving the value of what we can do with our current resources and then building on that foundation.”
5. Incorporating a Level of Abstraction for Enhanced Speed
Q: Why do you add a level of abstraction to the tools you use?
Adding a level of abstraction in the tools used can enhance team efficiency. By focusing on building models and leveraging pre-built components, teams can deliver results faster.
“Abstraction generally means we have to go elsewhere and get something that has been built and ready to go. Major cloud providers are usually the first place to go to.”
This approach allows the team to focus on core tasks and leverage external solutions for non-core functionalities, speeding up the development process. He added:”If we spend time trying to build a feature which is not our core strength, we’re basically losing time and money”
Final Thoughts
Philip’s insights offer a practical guide for AI leaders aiming to balance speed and efficiency in their projects. By prioritizing through cost-benefit analysis, balancing short-term and long-term goals, leveraging existing solutions, and building team capabilities, leaders can drive successful AI initiatives. His emphasis on abstraction and efficient tool usage further underscores the importance of streamlining processes to achieve faster results.
Get all the insights from the summit
Get to know the speakers
By Alexander Gorgin, Analyst at Vectice