LLM Engineering Intern

About Promoted.ai

Marketplace Optimization

Tech description:

We build unified streaming data and ML infrastructure for search, discovery, ads, and promotions similar to that used at social media companies and top marketplaces like DoorDash and Airbnb, which we license to other A16Z top 100 marketplaces and e-commerce apps.

We're ex-in-house-team Google and Facebook engineers. The technology is best practices self-service performance ad tech (Flink, click and conversion models, auto ML, multi-level retrieval, ranking, and auction) with the inefficiencies of a big company stripped out. For example, there is no separate "ads" system: all optimization and metrics run through unified search, feed, and ads data and ML systems. We replaced most ML signal and parameter experimentation with auto ML and maximizers because we move faster without launch reviews to quantify every incremental change. We own the entire stack: web and mobile eventing, data stream joining and aggregation, real-time ML training, retrieval, ranking, auction, manager, and reporting so that all our systems work nicely together without being blocked by others. We don't have a problem with the ads team cannibalizing the product to pull forward quarterly revenue to track estimates because we maximize the entire marketplace together: ads plus organic.

Our insight is that all commercial media can be better optimized as ads, even when that media not explicitly advertising. Our goal is to network together the marketplaces that we optimize with other media to build distributed ad words.

Job description:

Promoted.ai solves search relevance when it's the critical driver of business success: at large online marketplaces and scaled e-commerce companies. We combine the best of the last generation of ML with the latest in LLM technologies to increase revenue by 5-10% from dramatically better search, recommendations, and native ads.

We are hiring summer interns to rapidly experiment with generative AI and LLM tools. The ideal candidate is a computer science undergrad or graduate student at a top engineering school who loves to hack on new tools and techniques at scale. Ideally, you can work in our SF office, but remote in the USA is OK. We're a team of industry veterans. We've found that having a few junior people hacking on new experimental stuff can produce great insights. We will re-write your code if it works: we're looking for fast iterations and a fresh perspective. You should be highly self-motivated, responsible, and good at presenting your work.