(Senior) Analytics Engineer

Location: Los Angeles, in person (preferred) 
Salary range: $110.000–$165.000

TubeScience produces 8,000 original performance ads a month out of a 100,000 sq ft studio in Los Angeles. We are Meta's largest creative partner and AppLovin's number one. The ad library runs to 1.6 million pieces, each one tagged with what it actually earned against $2B a year in managed spend. As first-party creative-performance datasets go, there is not much else like it.

The role

You would build the data systems that connect ad performance to creative decisions. Our platform brings together data from advertising channels, creative production workflows and business operations. Your work makes that data reliable, understandable and useful in reports, internal products, and future analytics and machine learning applications.

It is hands-on across the whole data lifecycle: extracting data from APIs, maintaining dependable pipelines, modeling data in Snowflake, defining metrics, and delivering insights through Power BI and other tools. Much of the foundation already exists. You will learn how its parts fit together, make it more reliable and faster, and help shape what we build next.

We are open to hiring at Analytics Engineer or Senior Analytics Engineer level, depending on experience and scope of ownership.

Relevant experience matters here

We weigh directly relevant experience heavily. This platform carries real advertising data and feeds reporting the business runs on from the first week, so tell us plainly where your background maps: advertising data, warehouse modeling, pipelines and BI.

You would fit if

• You have at least 5 years in analytics engineering, data engineering or a closely related role, with meaningful ownership of production data systems.
• You have worked with advertising or marketing performance data, ideally pulling it directly from platforms such as Meta, TikTok, Google or Snapchat.
• You are strong in SQL and Python, and you can reason through the full path from a source API to a business-facing metric or report.
• You have designed data models and warehouse architecture that support changing business needs, including layered approaches such as medallion architecture where useful.
• You have built or maintained orchestrated pipelines and understand failure recovery, backfills, data quality and operational reliability.
• You have developed Power BI reports or semantic datasets and worked with stakeholders to resolve ambiguous metric definitions.
• You are comfortable with GitHub-based development, code review, and cloud infrastructure on AWS or GCP.
• You investigate problems independently, make practical engineering decisions, and explain the trade-offs to technical and business partners alike.

Useful, not required: Creative analytics, attribution or cross-platform performance measurement. Snowflake or Databricks performance optimization, including dynamic tables. Transformation tools such as Coalesce. Orchestration tools such as DBOS or Airflow. Datasets or features for internal applications, experimentation or machine learning. Fluency with AI-assisted development, while applying your own judgment to system design, correctness and production changes.

Probably not for you if

• You want someone to hand you specs.
• You want to work only on dashboards, or only on pipelines. This role covers the whole path.
• You want a greenfield rebuild. Much of the foundation exists, and improving it is part of the job.
• You want six months of design review before anything ships.

The problems

Advertising data from many platforms. Meta, TikTok, Google and Snapchat, each with its own API. Pagination, rate limits, historical backfills, schema changes, and reconciling what we store with what the platforms report. [Volume and freshness target.]

A platform that grows with the business. The path from external sources through ingestion, transformation and curated models to BI and application delivery, with architectural choices that still hold as data and use cases grow.

Metrics people trust. Performant Snowflake models that make advertising, creative, client and operational data consistent, with metric definitions agreed with stakeholders and documented.

Reporting that connects ads to creative. Power BI today, possibly migrating to a better BI tool. Datasets that link ad outcomes to creative strategy and production decisions.

Reliability without the toil. Investigating discrepancies and pipeline failures, adding checks and observability, and using agents to automate the mundane parts.

Feedback loops. Working with data, product, engineering and business partners to find analytics and machine learning approaches that feed ad performance back into creative development.

Stack

Snowflake, Python, SQL, Power BI, GitHub-based development and code review. If something in that list is wrong, you are the person who gets to say so.