As a Data Scientist Manager at Publicis Sapient, Christina Valore works at the intersection of data science and generative AI building, products that help creative teams produce images, videos, and content faster and more efficiently. Her work is fast-paced, experimental, but also practical in application. One of her most recent projects brings this practicality to life in a simple and relatable way. Her and her team built an internal tool for their client to accelerate the ideation of image drafts that can easily be incorporated into a campaign or ad assets.
Essentially the tool accepts a user prompt and generates up to four image drafts. These outputs are not intended to be as final deliverables but more as starting points that designers can clean up and adapt as needed. This tool significantly shortens the timeline from ideation, creative creation to launch so that creative teams can focus on refinement and storytelling instead of staring at a blank canvas. The core tool was developed in as little as six weeks. This is remarkably fast compared to traditional briefs and asset production timelines that can take months and is all the result of close collaboration across data science, engineering, and creative stakeholders with a willingness to ship minimal, useful functionality early.
In the fast paced, ever-changing landscape of AI, it can be challenging to stay on top of what’s next but it’s also what excites Christina about her work at Publicis Sapient. “Working on the bleeding edge of AI is exhilarating,” she says. “New capabilities appear frequently, and Publicis Sapient focuses on being ahead of the curve in how these tools can make work easier for our clients and us as employees.” She goes on to share an example of how quickly this space can change. “AI images are almost a thing of the past and now generative video is really emerging as the new frontier. Early models, for example, Veo 3.1, can produce short, rough clips in eight-second sequences that often need substantial polishing. However, the potential is enormous and something I’ve been keeping an eye on. The ongoing challenge is to transform those rough outputs into reliable, production-ready pipelines.” Staying current requires a mixture of reading and doing according to Christina. She and her team keep up with the latest scientific papers, then get hands-on with promising tools emerging in the marketplace. Within Publicis Sapient, a monthly data science all-hands highlights what different teams are building and shares breakthroughs. Additionally, the core data science team meets weekly for stand-ups to surface challenges, coordinate on change management, and align on long-term visions. In conjunction with these open forums, Christina leans into her background as a scientist and leverages the scientific framework when testing and evaluating new tools. Generative models are inherently unpredictable, so applying frameworks for measurement and control is essential. The team runs experiments, tracks outputs, and builds feedback loops so models improve and outputs become more reliable; thus, resulting in better generative tool outputs.
Working in an environment that is surrounded by constant change can present it’s challenges and Christina has experienced that firsthand. “As scientists, you can easily go down rabbit holes, so discipline is needed to keep experiments focused on product goals.” She goes on and says, “it is so important to ask for help. If you’re confused, speak up. Curiosity and humility accelerate learning, and you have to leave your insecurities at the door especially in this type of environment. We’re all learning and evolving together. Lastly, I’d say embrace difficult assignments. Being ‘thrown to the wolves’ early like leading projects before feeling ready, is one of the fastest ways to learn. Mistakes are part of growth and you shouldn’t shy away from that.”
