Machine Learning Scientist

Remotefull timeRemote

Apply on Employer Site

Machine Learning Scientist at Blueroseresearch · LumiJobs

mid

$140k–$190k

H1B: no

Blueroseresearch The Machine Learning Scientist at Blue Rose Research builds and maintains modeling pipelines and develops models to support electoral forecasting and polling analysis. The role involves working with messy political data and translating statistical outputs into actionable guidance for campaigns and organizations.

Responsibilities

We're hiring a Machine Learning Scientist to join our Scaled Polling & Forecasting Team, the group responsible for Blue Rose's flagship electoral work. Our team builds and maintains models that help progressive organizations understand the electoral landscape and make informed decisions. We track public opinion, forecast races across the ballot, develop resource allocation tools, and deliver analysis that campaigns and allied organizations rely on to set strategy.
Our problems are not well suited to off-the-shelf ML solutions: we work with messy, evolving political data and constantly pressure-test our accuracy against real election outcomes. The role spans the full modeling lifecycle from raw survey data to predictive scores and requires both rigorous statistical thinking and the judgment to know when a number is real versus an artifact.
Specifically, you'll:
Build and maintain end-to-end modeling pipelines, improving reliability, data quality checks, and model diagnostics to increase confidence in our outputs.
Develop models and perform feature engineering, contributing new ideas to improve our polling and survey infrastructure. You'll think creatively about the right ways to evaluate model performance in contexts where standard metrics don't apply cleanly.
Models include Identity Resolution, Record Linkage, and Forecasting models.
Translate statistical outputs into actionable guidance for internal stakeholders, connecting model outputs to the strategic questions campaigns and organizations care about.
Scope technical needs and ship solutions alongside engineers, statisticians, and political analysts.
Build subject matter context, developing deep knowledge of the political landscape to inform modeling decisions and catch anomalies.

Qualification

ExperienceTechnicalCollaborativeA strong communicatorCuriousA good teammate

Preferred

Experience: 4+ years of professional experience in machine learning engineering or research science with applied statistics and predictive modeling OR equivalent depth from a PhD program. We're open to both paths. What matters is strong instincts for evaluating model performance, diagnosing bugs, and making principled judgment calls under uncertainty.
Technical: Confident Python programmer who can navigate a codebase and implement extensions. Strong data-wrangling experience, ideally including SQL and relational databases. We are particularly interested in ML expertise in Identity Resolution, Matching / Record Linkage, Survey and Voterfile Modeling, and/or scaled Bayesian Modeling (numpyro/JAX), but all talented ML researchers should apply.
Collaborative: You thrive in multi-disciplinary teams working alongside engineers, statisticians, and political experts, and you're excited to partner with less technical stakeholders and see your work impact real-world decisions.
A strong communicator: You can translate between statistical and substantive thinking, explaining what a model is saying (and not saying) to people who think in terms of strategy, not standard errors.
Curious: You're eager to engage with the wider progressive political ecosystem and develop domain knowledge alongside your technical skills.
A good teammate: You're a kind person who contributes to a warm working environment.
This is a remote or in-person optional position. We have an office in New York City and a number of team members who work in-person regularly.
Our teams work on mostly East Coast hours.