Hi, I'm Jay
Product leader and AI builder
About me
I'm passionate about creating great products that solve hard problems.
I believe the best outcomes result from open collaboration — between product and engineering, of course, but also via frequent interactions with clients and prospects, go-to-market teams, the C-suite, and anyone else with curiosity and a drive to help create best-in-class software solutions. Particularly important is the ability to translate in-the-weeds concepts into layman's terms for non-technical audiences, and vice versa.
My career has spanned everything from Series A-stage tech startups to sprawling, 150,000-person organizations. The startups I've joined were acquired (twice!) for over $1.2 billion collectively. My expertise is concentrated in launching large-scale technical products. As the head of brand safety product management, I led an international team of PMs to build best-in-class contextual data products that generated millions in revenue within their first three years. These included API-first viewability and fraud prevention integrations with the largest buying platforms in the digital advertising ecosystem, as well as highly bespoke and first-in-market measurement integrations with marquee social platforms including Facebook, Instagram, Snapchat, and Pinterest.
I like getting my hands dirty: I've leveraged my technical curiosity to teach myself multiple languages -- Ruby on Rails, Python, and JavaScript -- and launched a wide variety of apps (see below projects), and I'm comfortable using a broad variety of cloud services, tools, and frameworks (AWS, Hugging Face, Streamlit, and Retool, for example) in a wide swathe of applications ranging from audio analysis to model fine-tuning. Lately I've been building and open-sourcing LLM-powered tools to summarize podcasts and analyze transcript accuracy.
I'm always up for a chat!
Projects
Saving you from mild discomfort at cocktail parties.
Evaluating the semantic accuracy of automatic speech recognition (ASR) transcripts.
An interactive financial portfolio of Tether.
Graphing the Federal Reserve's interest rate projections over time.
The definitive resource for tracking United States Supreme Court justices' public engagements. (No longer maintained.)
Recent Substack posts
When political polling and betting odds diverge
I've created a new tool to correlate pairs of time-series datasets. So now you can measure just how much Thursday's debacle split the presidential polling average from the PredictIt odds.
A smarter way to measure transcript accuracy
TLDR: I've built a Transcript Accuracy Analyzer app to evaluate accuracy in speech-to-text ASR models. (Warning: this gets pretty in the weeds.)
Careless Whisper (Usage Can Blow Out Your API Limits)
Fun new experiments with cost savings on OpenAI.
Get in touch
