Published Dec 17, 2024

SE Radio 647: Praveen Gujar on Gen AI for Digital Ad Tech Platforms

Praveen Gujar, Director of Product at LinkedIn, delves into the revolutionary impact of generative AI on digital advertising, highlighting advancements in ad content creation, audience targeting, and campaign management, while emphasizing strategic AI integration and privacy compliance to boost efficiency and ROI.
Episode Highlights
Software Engineering Radio - the podcast for professional software developers logo

Popular Clips

Questions from this episode

Episode Highlights

  • AI Teams

    AI integration within product development teams is reshaping organizational dynamics. explains that modern teams now include AI engineers and data scientists alongside traditional roles, fostering a more cohesive approach to product development 1. This integration enhances AI proficiency across the team, enabling engineers to better understand and utilize AI capabilities. emphasizes the importance of AI proficiency, noting, "Even if they may not be building the AI models, they are proficient in a way to train an AI model, do prompt engineering if needed as well and use the AI model in the right way."

    Even if they may not be building the AI models, they are proficient in a way to train an AI model, do prompt engineering if needed as well and use the AI model in the right way.

    ---

    This shift towards AI-integrated teams is crucial for developing AI-first products and leveraging AI's full potential 2.

       

    Model Strategies

    Selecting the right AI models is critical for optimizing ad platforms. outlines the importance of considering factors like data type and computational resources when choosing models 3. For instance, large models are suited for audience segmentation, while smaller models suffice for tasks like ad copy generation. Gujar highlights the trade-offs involved, stating, "Smaller models are trained on much fewer parameters than a large language model, they take less computational capacity from a performance perspective."

    Smaller models are trained on much fewer parameters than a large language model, they take less computational capacity from a performance perspective.

    ---

    This strategic model selection ensures efficiency and transparency in campaign management, balancing performance with resource demands 4.

Related Episodes