Interview Questions for Priscilla Alexander, Vice President of Engineering at ArthurAI

In today's digital age, businesses are increasingly turning to machine learning (ML) for decision-making processes. However, understanding the reasoning behind these decisions is crucial for trust and compliance. This is where ArthurAI comes in, a platform designed to provide explainability for ML models, offering a strategic advantage for companies embracing AI.

The Challenges of Building ML Software

Building working ML software presents numerous challenges, including the non-deterministic nature of decision-making and the need for active monitoring to ensure accuracy and compliance. These challenges can be daunting for businesses, leading some to consider partnering with specialized AI companies like ArthurAI.

The Arthur Platform: A Game Changer

ArthurAI's flagship product, the Arthur platform, has made a significant impact in various industries. For instance, it helped a team at Harvard's Dumbarton Oaks Institute improve the accuracy of a computer vision model used for analysing photographs by identifying areas where the training examples were insufficient.

The platform operationalizes explainability for every inference and across all model decisions, a critical aspect from a regulatory perspective in financial services, where decisions like loan recommendations or investment decisions need to be transparent. Moreover, the Arthur platform can run anywhere models run, whether on AWS, GCP, or in a data centre with no internet access.

Filling a Gap in the Market

ArthurAI aims to fill a gap in the market by taking care of the operational production monitoring of ML models, allowing teams to focus on their business-critical challenges. The platform's explainability and bias mitigation techniques are bleeding edge, but there is still a lot more research to be done in these fields.

Overcoming Barriers to AI Adoption

Adoption of AI is often hindered by several factors, including the availability of data, scarce and expensive talent, uneducated business leaders, and concerns about regulatory compliance. ArthurAI addresses these challenges by providing specialized AI observability tools and expertise, making AI systems safer, fairer, and more transparent.

The Future of AI with ArthurAI

ArthurAI's software can instrument a model in a customer's model serving architecture in just a few minutes. The platform finds disparate impact by computing the difference in outcomes across subpopulations, scaling this comparison to hundreds or thousands of subpopulations. It can also detect unfairness in a model and support techniques to mitigate the outcomes, although humans are needed to decide which technique to use due to the tradeoffs involved.

The startup, co-founded by Priscilla Alexander, who previously led teams that built ML applications at Capital One, works with industries focused on AI-driven risk management, including finance, technology, health care, and manufacturing. By collaborating with ArthurAI, these industries can address challenges such as AI model monitoring, bias detection, fairness, and regulatory compliance, ensuring their AI systems are not only effective but also transparent and fair.

Interview Questions for Priscilla Alexander, Vice President of Engineering at ArthurAI