Bay Area]

TubeScience-Labs•Los Angeles, CA

17h•$120,000 - $200,000•Onsite

About The Position

TubeScience Labs is seeking a Staff / Principal Software Engineer to join a small, AI-empowered, senior team. The company's mission is to create trusted, scalable, and self-improving AI systems for the largest performance-based paid social creative video company globally. TubeScience is a major creative partner for Meta and AppLovin, producing a high volume of original ads monthly from a large studio, leveraging a vast library of performance ads and significant annual managed ad spend. This provides access to a rich first-party creative-performance dataset. The Labs division transforms this data and expertise into frontier AI tools. The role involves working across all frontier and open-weight models for discovery, prototyping, specification, building, evaluation, and deployment. The engineer will own the architecture direction for their area, including what gets built, the order of development, release cadence, and what is excluded. While agents will handle coding, the engineer remains accountable for all shipped code. The team operates with minimal hierarchy, providing direct access to users and immediate feedback. The company values directly relevant experience, with roles offering real budgets, footage, and operators from day one, minimizing the need for domain learning. Two product areas are available: Digital Production, focusing on a full-stack system to transform prototypes into finished videos by shifting work from human editing to AI-assembled variants, requiring understanding of editing and generative workflows. Media, focusing on systems for post-ad creation and approval, including deployment to ad platforms, spend allocation, experimentation, measurement, and reporting, requiring depth in advertising systems like media buying, ad-platform automation, spend allocation, bidding, measurement, attribution, experimentation, or ranking/recommendation.

Requirements

Held Staff or Principal scope.
Set technical direction across multiple systems or teams, and be able to name specific calls and their outcomes.
Depth in the domain of the area applied for (Digital Production: video editing, media processing, or generative media pipelines; Media Effectiveness: advertising, media, or performance-data systems).
Built a platform from zero, made foundational calls, and lived with them.
Put AI into production beyond chat (e.g., vendor APIs, agents, tool calling, evaluation, permissions, cost control, running against real users).
Go a layer or two below the abstraction by instinct.
Worked with at least one systems-level language like C, C++, Rust or Zig.
Used Python and TypeScript in production, both ends of the stack, at a level where you have opinions about trade-offs.
Shipped to real customers and handled feedback.

Nice To Haves

Experience with FFmpeg, CDNs, ingestion or editing tools.
Dealt with model routing at scale.
Spent time inside an adtech company, agency or production studio.

Responsibilities

Own the architecture direction for your area: what gets built, in what order, and why.
Define the release cadence and what does not get built.
Direct a fleet of agents for coding tasks.
Ensure accountability for all shipped code.
Work across all frontier and open-weight models for every phase of the job from discovery to prototyping, to specs and building, and to evals and deployment.
Set the technical direction for your area, ship code every week, and decide what does not get built.
For Digital Production: Shift work from human editing toward AI-assembled variants, working with AI tools and traditional editing software like Adobe Photoshop and Premiere Pro.
For Media: Develop systems for deploying creative to ad platforms, efficiently allocating spend, running experiments, and measuring and reporting performance against client goals.
Build robust media pipelines handling many assets a day through ingest, transform and render.
Create durable workflows for long-running jobs that survive restarts, partial failure, and changes while in flight.
Implement model routing and spend management, tracking permissions and costs per team, and choosing the right model per job without human intervention.
Develop agents that make real decisions, including planning systems that commit budget, book people, and set schedules, trusted by operators.
Ensure coherence across several products sharing data, users, and infrastructure by holding the architecture and standards.
Enable shipping in days, going from suggestion to production rapidly.