Skip to content
View aranculleton's full-sized avatar
1 follower · 3 following

Achievements

Achievement: Pull Sharkx3Achievement: YOLO

Achievements

Achievement: Pull Sharkx3Achievement: YOLO

Block or report aranculleton

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
aranculleton/README.md

Aran Culleton

Data Science Consultant at Xantura (Newton Europe) · ex-British Airways and BBC · Toronto

I build models that change real decisions, and the pipelines that keep them working in production. My background is financial economics, and I've worked on risk, pricing and experimentation in aviation, media, fintech and the public sector. I like owning the whole job, from the raw data to the conversation with the people who use the results.

  • Xantura (Newton Europe): I build risk-scoring models that rank 40k+ accounts across 1.5M+ sensitive public-sector records, so frontline teams know where to act first.
  • British Airways: I built pricing and market-share forecasts from 200M+ ticket records (~5% MAPE) and presented them to American Airlines and Japan Airlines.
  • BBC: I ran A/B tests on BBC Sounds (5M+ monthly users). The winning variant lifted click-through by 6%.
  • Nova Money (fintech startup): I worked directly with the CEO on onboarding experiments that helped triple monthly sign-ups, from ~1.5k to ~4.5k.

Stack: Python · SQL · scikit-learn · Airflow · dbt · Azure Data Factory · Snowflake · Databricks · AWS · Power BI
Certified: Databricks Data Engineer Associate · AWS Machine Learning Engineer Associate

Projects

  • last-job-search: an AI agent system that runs a whole job search from one page. It scores thousands of postings, writes fact-checked tailored applications and drafts the networking messages.
  • openroles: a Python library and MCP server. Give it a company name and it pulls every open role straight from that company's hiring system (Greenhouse, Lever, Ashby, Workday and more). On PyPI.
  • CaseSignal: a credit-risk prototype inspired by my work at Xantura. It ranks accounts by risk, and adding auditable AI signals from case notes raises precision in the top 10% by 14%.
  • CopulAX (contributor): a JAX library for copula modelling and probability distributions. I'm adding support for univariate distributions, with tests for the edge cases.

Away from data

I love to cook. When I'm not in the kitchen, I'm usually travelling or watching films, and I log every film on Letterboxd.

LinkedIn

Pinned Loading

  1. last-job-search last-job-search Public template

    An AI agent system on Claude Code that runs your whole job search: finds and scores roles, creates and tailors your resume for every role, finds who to message on LinkedIn, and reminds you to follo…

    Python 7

  2. openroles openroles Public

    Every open job role, straight from an employer’s own hiring system (Greenhouse, Lever, Ashby, Workday and more). No API key, no dependencies. CLI, Python API and MCP server.

    Python 1