Seneca Widvey, Director of Knowledge Science, Audacy, Inc
My journey into information science didn’t begin in pc science or statistics—it started in psychology. I used to be captivated by cognitive science and neuroscience, particularly the methods we will use computational fashions to raised perceive human conduct. As that curiosity grew, I noticed I wanted a deeper mathematical basis to reply the sorts of questions I used to be asking. That realization led me to pursue a second diploma in utilized arithmetic, and later a grasp’s in arithmetic and statistics at Georgetown College, the place I centered on information mining, stochastic processes, and machine studying.
After graduate faculty, I joined IBM, the place I labored on making use of machine studying to complicated, high-impact challenges in authorities and healthcare. My initiatives included fraud detection and danger modeling, areas the place data-driven decision-making can have real-world penalties. That have taught me quite a bit concerning the messy realities of deploying machine studying in massive, multifaceted organizations.
My transition into media and leisure got here once I joined Audacy, a nationwide chief in radio and podcasting. What drew me in was the possibility to use superior information science to a dynamic business with wealthy, real-time information. From personalization and audio-transcription fashions to forecasting, retrieval-augmented technology (RAG), and large-scale experimentation, the probabilities felt infinite. I’ve since grown into the position of Director of Knowledge Science, the place I lead a cross-functional staff of knowledge scientists and ML engineers centered on driving innovation aligned with Audacy’s broader enterprise technique.
Aligning Knowledge Science with Technique and ROI
To make sure each information science initiative delivers measurable worth, I depend on an idea borrowed from training: scaffolding. As scaffolding helps learners by complicated materials, structured frameworks assist information science groups navigate uncertainty and keep centered on strategic priorities.
At Audacy, we function in Scrum sprints, permitting our information science work to align tightly with engineering and product. This shared cadence strengthens communication and helps us combine fashions easily into bigger platforms. We comply with the information science lifecycle and MLOps frameworks rigorously, from preliminary enterprise understanding by information exploration, modeling, deployment, and monitoring.
What makes our experimentation framework profitable is the connection between technical rigor and enterprise context. It’s not nearly proving one thing works—it’s about understanding why it really works, who it advantages, and the way it suits into our firm’s bigger objectives 
One key focus is information high quality. Even probably the most subtle fashions can’t carry out properly on messy or inconsistent information, so we make investments early in making certain our information is dependable and accessible. Earlier than any modeling begins, we align with stakeholders on success metrics and make clear the enterprise selections we goal to help. For extra exploratory efforts, we use versatile frameworks like OSEMN or Kanban. However as we transfer towards manufacturing, construction, governance, and automation take middle stage. This mixture of flexibility and rigor helps us ship influence transparently and at scale.
Driving Change by Communication and Belief
One of many greatest hurdles I’ve confronted is translating complicated insights into one thing that resonates with non-technical stakeholders. With a background in utilized math, I really like diving into the main points of a mannequin, however I’ve discovered that technical precision alone doesn’t drive selections. What issues extra is answering the query: So what?
If a mannequin improves accuracy by 5 %, what does that imply for the enterprise? Will it improve income? Improve buyer satisfaction? Enhance operational effectivity? After we can clearly articulate that influence, it turns into a lot simpler for leaders to take motion.
Constructing belief throughout departments, from product and engineering to editorial and govt management, requires delivering well timed, actionable, and straight related insights to enterprise outcomes. At Audacy, we additionally foster a tradition of analytics by workshops, transparency, and shared success metrics. These practices assist make information science not only a help operate however a strategic driver.
Establishing a Tradition of Experimentation
Once I joined Audacy, considered one of my first objectives was to determine a proper A/B testing program. On the time, product selections weren’t experimentally pushed. I partnered with the Product staff to construct a framework that supported structured, data-driven experimentation.
Good experimentation begins with clear hypotheses. For example, if we’re testing a brand new app structure, we don’t simply ask whether or not customers prefer it—we ask whether or not it improves listening hours or enhances engagement, and we outline particular metrics to measure these outcomes. We additionally monitor counter-metrics to catch any unintended penalties.
A/B testing is particularly highly effective when utilized to machine studying. Earlier than launching a advice system, for instance, we take a look at completely different fashions in managed environments, evaluating metrics like click-through charges and time on the platform. We usually begin small, testing on restricted consumer teams, to handle danger and collect insights earlier than scaling up.
What makes our experimentation framework profitable is the connection between technical rigor and enterprise context. It’s not nearly proving one thing works—it’s about understanding why it really works, who it advantages, and the way it suits into our firm’s bigger objectives.
Main with Objective in a Quick-Paced Business
If I may supply one piece of recommendation to information science professionals trying to lead in consumer- centered industries like media and leisure, it will be this: put money into your folks.
At Audacy, I’m lucky to guide a staff of extremely gifted information scientists and machine studying engineers, every bringing distinctive strengths and views. Common knowledge- sharing periods are among the many greatest methods to harness that variety. Whether or not exploring new modeling strategies, reviewing educational analysis, or debriefing a deployment, these periods promote curiosity, collaboration, and collective development.
Management, particularly in a fast-moving subject like ours, requires energetic engagement. It means understanding what your staff is constructing and what they should succeed—whether or not that’s resolving a technical blocker, securing sources, or connecting their work to enterprise technique. I embrace the Servant-Chief mannequin: my position is to empower others to do their greatest work, problem one another constructively, and proceed evolving as professionals.
