Data Science Intern (Python + pandas)

Recruiters on MailerMen reject roughly seventy percent of the applications they receive, and nobody here can currently say which signals predict that rejection. Finding out is your project.

Three months, one question

You will build a candidate to job relevance model from scratch. That means pulling application history, engineering features from resume text, job description text and behavioural signals such as time to apply, then testing whether a gradient boosted model beats the keyword matching we run today. Baselines matter. If TF-IDF and logistic regression win, we ship that instead.

How we work

This role is open worldwide and runs asynchronously. Each week you write a short memo describing what you tried, what failed and what the numbers said, and a senior data scientist replies in writing with questions. Experiments live in a shared repository with seeds fixed, because we care that a result can be reproduced next month.

Requirements in plain language

You need real comfort with Python and pandas, an understanding of train and test splits that goes past calling the function, and enough statistics to know when a two point lift is noise. No professional experience expected. Coursework, a Kaggle notebook or a personal project you can talk through in detail is exactly the right evidence.