Note: The job is a remote job and is open to candidates in USA. Launch Potato is hiring a Staff Data Scientist to build and operate models that match auto insurance shoppers with carriers and guide media buying decisions. The role owns the data science workflow end to end, from problem framing and data discovery through model deployment and monitoring, with scope expanding to other verticals over time.

Responsibilities

• You'll build the models that decide which carrier each auto insurance shopper sees on our click wall
• The goal is to send every user to the carrier most likely to convert them, so carriers see our traffic convert and raise their bids
• You'll also own the buying models behind our media spend, which are held to ROAS
• Carrier conversion data (quotes, policies and acquisition cost) is now coming back to us, and you'll be the one who puts it to work in production
• You frame problems directly with the business, find your own data, build and validate the model, and stay with it through deployment and monitoring
• An ML engineer works alongside you on the production side
• Auto insurance takes most of your time to start, and scope broadens to other verticals over time
• By day 60, you run our insurance marketplace media buying on your own
• Click wall serving that matches each user to the right carrier
• Carrier conversion data running in production models
• Automated marketplace bidding taken from proof of concept to scale
• A clear approach for repeat users and sparse sessions
• Production models that hold up through drift and changing objectives
• Output the business trusts and uses without rework

Skills

• 5+ years of hands-on applied data science with measurable business impact
• Your models drove paid media decisions: spend, bidding or ad serving. Attribution and reporting alone don't qualify
• You built what's on your resume and can explain it down to the loss function and architecture choice, neural network fundamentals included
• You see how today's bids shape tomorrow's training data, and you've handled late labels, sparse features or drift
• You own models end to end: you find your own data, build through deployment and monitoring, and bring a point of view
• Expert Python and SQL
• Real ad budget spent in live auctions
• Large-scale ranking or recommendation systems
• Explaining tradeoffs to non-technical partners in plain terms
• Designing for auction feedback loops, selection bias and exploration
• Hands-on neural network work, such as embeddings of text, creative or user behavior
• AWS for data science: ECS, S3, Athena, SageMaker
• CI/CD deployment

Benefits

• Remote-first work arrangement
• Profit-sharing bonus

Company Overview