Content Louis is Following

With the rise of language models and GenAI, entering an AI team has become easier, as business solutions can now be built using existing large models through inference. Previously, data scientists needed skills in statistics, engineering, and linear algebra to train custom models. Now, significant results can be achieved by running inferences on pre-trained models without those skills. In your opinion, what are the main arguments for keeping inference capabilities within an existing team of data scientists rather than outsourcing them?

FeedbackAI & Machine Learning
52 views

As a leader of a company's internal AI organization, what techniques you use to divulge your team's capabilities and services across the organization?

Peer InsightsFeedback
Louis BruderDirector of Data at FootLocker5 months ago
We run a Digital Council across the company to support what capabilities the organization has and how they are evolving.   We also run for 5 years a whole area of our intranet called *one Voice on Digital* that we use to ...read more
1 Reply
Read More Comments
264 viewscircle icon4 Comments

What metrics do you have in place to measure the bias, explainability and transparency of AI in your data and analytics ecosystems?

Disruptive & Emerging TechnologiesData & Analytics
Louis BruderDirector of Data at FootLocker7 months ago
In my case, this varies by the type of AI model in question. Overall, bias is tracked through a metric called Equal Opportunity Difference to ensure fair treatment across groups. Explainability is obtained through Shap values. ...read more
280 viewscircle icon1 Comment
Louis Bruder, Director of Data at FootLocker at FootLocker | Gartner Peer Community