AI Audio Content Intelligence Analyst

Clara Osei

Freelance · Mid-level · Los_Angeles

This role is no longer accepting applications

Similar open roles you can apply to now:

About the role

We make podcasts. A lot of podcasts. Currently 52 active shows, 13 million downloads a month across the network. We know which shows do well. We know which episodes get shared. We know roughly when people stop listening, because our hosting platform tells us. What we do not know is why any of it works, and that is frankly embarrassing for a company that talks about being data-driven in every investor deck we have ever produced. We have transcripts. We have engagement data. We have guest metadata and topic tags and listener demographic estimates from our ad network. We have a data analyst who is very good at making charts but has reached the limits of what Excel and vibes can tell us. We are looking for someone to run a focused four-week project: take what we have, apply modern NLP and pattern analysis to it, and give us a model — or at minimum a clear framework — for predicting which episodes will over-perform before they go live. We are not expecting a production system. We are expecting rigorous analysis, clear documentation, and a set of findings we can act on. If you find something interesting we did not ask about, we want to hear it. This is the kind of project that could become ongoing work if the output is good.

Key Deliverables

List the expected deliverables for this project.

Audit and clean the transcript and engagement dataset for the project scope — 18 months of episode data across 12 flagship shows
Build a feature set from transcript content, guest metadata, and episode structural characteristics
Develop and evaluate predictive models or scoring frameworks for episode performance
Produce a written analysis report and a Tableau dashboard presenting key findings for the editorial and commercial teams
Present findings in a 60-minute session with the leadership team at the end of the engagement

Requirements

Technical stack needed for this mission.

3+ years of data analysis or data science with strong NLP or text analytics experience
Python — comfortable with Pandas, text preprocessing, and basic ML workflows without heavy scaffolding
SQL for querying our episode, engagement, and subscriber data from our analytics warehouse
Statistics: the ability to distinguish a real pattern from noise in a messy real-world dataset and communicate that distinction clearly
RAG or embedding-based analysis experience — being able to build a semantic similarity layer over transcript data would be very useful
Tableau or equivalent for producing analysis outputs that a non-technical podcast producer can look at and understand