Applied AI Engineer

Our client is a seed-stage AI company, backed by a top-tier venture firm, building the ad layer for the AI era: interactive brand experiences embedded natively into apps and websites. Its ads are generated dynamically in real time and personalized to every single user, so every impression is hyper-relevant.

The recommendation stack is already live, serving ads across millions of daily interactions, reaching tens of millions of users, with millions in annualized ad spend. Dozens of AI-generated playable mini-game and video ads run in production inside major AI apps, built on the company's own playable engine and editor, and its ad SDK (Swift, Kotlin, React, React Native) is live in top App Store and Google Play apps. The team is small and based in San Francisco, with multiple prior exits and experience from leading tech and finance companies.

The Applied AI Engineer owns the engine that makes real-time, personalized ads possible: agents that generate playable mini-games, short-form video, and every variant in between on the fly, then learn from how they perform across millions of daily interactions and millions in annualized ad spend. This is a full-stack applied AI role spanning model selection, agent orchestration, and production infrastructure, with direct business impact at every layer.

The architecture isn't defined yet, and this hire helps define the approach. The agents plug into a recommendation stack that is already serving ads in production, and every ad they create is served to real users and measured against real advertiser dollars. The stack runs on GCP, Redis, MongoDB, and Temporal. The role comes with ownership, equity, and a direct line to what ships.

• Playable generation agents: Agents that turn an advertiser or app brief into a working, on-brand HTML5 playable ad.
• Automated QA: Checks that catch broken playable builds before a human does.
• Video generation pipelines: Multimodal pipelines that produce short-form video ads and variants across formats, hooks, and styles using frontier image and video models.
• Real-time personalization loop: Connect generation to the recommendation and ranking engine so each user gets creative made for them, read what wins (engagement, CTR, installs, ROAS), and feed it back so every impression gets more relevant.
• Agent infrastructure: Orchestration, evals, and reliability so all of this runs in production at scale, not just in a demo.

You inherit a working system with real traffic, not a blank slate:

• A recommendation stack already serving ads across millions of daily interactions, reaching tens of millions of users with millions of dollars in annualized ad spend
• Dozens of AI-generated playable mini-game and video ads running in production inside major AI apps, built on the company's own playable engine and editor
• A proprietary ad SDK (Swift, Kotlin, React, React Native), live in top App Store and Google Play apps

1. Strong applied AI experience: has built and shipped an LLM, multimodal, or agentic system in production, beyond simple API wrappers.
2. Backend depth across infra, data, and serving end-to-end, not just the model layer (GCP, Redis, MongoDB, Temporal).
3. Comfort with ambiguity: the architecture isn't defined yet, and this hire helps define the approach.
4. Bias toward shipping, suited to an early-stage company that moves fast and figures things out.

• Location: San Francisco, CA (in-person)
• Employment: Full-time
• Experience: 4+ years
• Salary: $200,000 – $250,000 USD per year, plus equity
• Visa Sponsorship: Not available

1. Interview with the CTO
2. Take-home assignment
3. Work trial
4. Offer