Product Portfolio

Allrecipes Skill

Want to see it in action? Ask Alexa to open “Allrecipes”

I was the lead PM in charge of re-launching Allrecipes’ Alexa skill from scratch in Fall 2019. The new skill effectively turns Amazon Alexa into a cooking assistant. Aspiring home chefs can ask Alexa for a new recipe and then receive interactive, hands-free guidance to prepare the dish in real time.

As someone OBSESSED with food, I loved the opportunity to launch a product like this. The skill was a massive undertaking for the 8-week timeline we had and only 2 developers. Over 30 thousand users interacted with the skill in the first week of launching! Working with the Allrecipes editorial team allowed us to execute a killer go-to-market strategy.

A major component of this project was metrics tracking. We decided to incorporate a service called Dashbot to help us track user behavior. Since launch, I have been sorting through the data to pull out the key metrics that will act as indicators to product success in the long term. I am in the process of setting up a funnel to see the percentage of users who end up cooking recipes. Then, I hope to set up a framework for A/B testing, since this is not available out-of-the box with Dashbot and would be well worth the effort to ensure that we are moving toward product-market fit!

Bixby Home Value Capsule

In 2018, I was working at Voiceter when Samsung announced that they would be releasing Bixby Developer tools. Anyone could add functionality to Bixby, Samsung’s personal assistant. Once word had reached Voiceter, we knew we had to build something! After speaking with the Samsung team, we determined that we could reach the most new users with a home price estimator. Shortly after, I started working on what would turn into my favorite product at Voiceter!

Despite countless challenges, we submitted the Bixby Marketplace’s first real estate capsule and earned Bixby Premier Developer status in only 6 weeks. We iterated every few days, saw what worked (and didn’t work), and adjusted our plan accordingly. I learned a ton in just a few weeks. The whole experience made me a better PM and reminded me of the power of rapid iteration.

Algolia Search Overhaul

During my time at Influenster, the Amazon CloudSearch system that we used was not aging gracefully. Looking for alternatives, I researched Algolia and presented my findings to our CTO. Despite being more expensive, we determined that creating a new search engine with Algolia would create the best search experience in the long run. Plus, the cost of the existing technical debt indicated something desperately needed to be done.

I worked with our UX designer to come up with an all-new user experience and designed the system that would sync our database with Algolia. This was important so that the search results would be up-to-date. Then I worked with 4 engineers to finish building the new search and record sync systems, making it faster, more reliable, and scalable. Overall, it was a huge improvement over the old search system!

Recommendation Engine

Influenster measured user engagement by tracking quarterly reviews. To improve this metric, we launched a product recommendation system to suggest products that users could review.

For the initial MVP, a section of the homepage showed products with a CTA to review. Simple, deterministic rules initially governed what products users saw. For example, if a user answered a question on a product, they might know enough about the product to write a review.

Then we really started to have fun with recommendations. We needed to get users to click on recommended products and write reviews! I designed several machine learning prototypes. After experimenting with which ones would accurately predict what product users reviewed, we landed on an “Alternating Least Squares” model. I was working to scale this model to handle over 4 trillion user/product combinations before I left Influenster to join Voiceter Pro.