pathfol.com

For data people

A data analyst portfolio that starts with the question, not the notebook.

Hiring managers rarely open a notebook. They read what you were asked, what you found, and what the business did next. Upload a CV, add GitHub if you use it, and write those three lines for each project.

Create your profileOpen the example

Example

What the finished page looks like.

A made-up staff data scientist, shown in the Orbit template. Priya Nair is not a real person.

Orbit · London

Priya Nair

Staff data scientist on fraud at Monzo

16+
years experience
1
projects shipped
2
companies
2
milestones
Priya Nair

Transmission

Priya works where a model meets a ledger. Research engineering at DeepMind in London, then the fraud and spending systems behind a Monzo card.

Stations

Every body in this system.

  1. 01 · Education

    2010-2014

    BSc Mathematics

    University of Warwick · Coventry

    Mathematics at Warwick, with the third year on probability and statistical modelling. Dissertation on time-series forecasts.

    • First-class honours
    StatisticsProbabilityUniversity of Warwick ↗

  2. 02 · Education

    2014-2015

    MSc Computing

    Imperial College London · London

    A one-year master’s in the Department of Computing, weighted toward machine learning. Thesis on evaluating sequence models when the labels arrive late.

    • Distinction
    Machine learningPythonImperial College London ↗

  3. 03 · Work

    2015-2020

    Research Engineer

    DeepMind · London

    Research engineer in the London lab. Built dataset pipelines and evaluation harnesses for applied machine-learning projects, including the science group’s public releases.

    • Owned evaluation jobs that compared model checkpoints against held-out sets
    • Supported dataset prep for research the London lab published
    EvaluationPythonData pipelinesGoogle DeepMind ↗

  4. 04 · Course

    2017-2017

    Practical Deep Learning for Coders

    fast.ai · Remote

    The fast.ai course, taken while at DeepMind, to keep a practitioner’s loop next to the lab’s research code.

    Deep learningPyTorchfast.ai ↗

  5. 05 · Achievement

    2019-2019

    Shared evaluation harness

    DeepMind · London

    The lab adopted her evaluation runner as the default for one research group: same splits, same metrics, a report a paper could cite internally.

    • Became the default runner for the group’s weekly results
    EvaluationGoogle DeepMind ↗

  6. 06 · Work · Now

    2020-Now

    Staff Data Scientist

    Monzo · London

    Staff data scientist on financial crime. Models that sit on the card payment path, plus the features behind Monzo Trends.

    • Leads the model that scores card payments for fraud before they settle
    • Partnered with product on the category totals in Monzo Trends
    FraudProduct analyticsSQLMonzo ↗

  7. 07 · Achievement

    2022-2022

    Trends category model

    Monzo · London

    Shipped a clearer merchant-category model for Monzo Trends, so a coffee and a grocery stop land in the buckets people expect.

    • Reduced ‘general’ as a dumping ground for known merchants
    ClassificationMonzo ↗

  8. 08 · Project

    2023-2024

    Borough Spend

    Independent · London

    A personal notebook of London spending patterns drawn from Office for National Statistics releases. No bank data. Public tables only.

    • Charts family spending by region from the ONS Living Costs release
    VisualizationPublic dataOffice for National Statistics ↗

Constellation

PythonEvaluationStatisticsProbabilityMachine learningData pipelinesDeep learningPyTorchFraudProduct analytics

Priya Nair

What belongs on it

Five things a reviewer looks for.

  1. 01

    The question behind each project

    “Why did repeat orders drop in March?” is a better title than “Sales analysis”. It tells the reader you worked on a real decision.

  2. 02

    What you found, in one sentence

    Lead with the finding. The method comes after, and only as much of it as a non-analyst could follow.

  3. 03

    What happened next

    A changed price, a cancelled campaign, a dashboard a team still opens. Analysis that led to nothing is hard to tell apart from homework.

  4. 04

    Tools, tied to the work

    SQL, Python, dbt, Tableau, each next to the role or project where you used it. That answers the screening question before it is asked.

  5. 05

    One public example

    A repository, a published dashboard, or a write-up. If your real work is confidential, a small project on public data shows the same habits.

  6. Leave out

    • The Titanic and Iris datasets. Reviewers have seen them hundreds of times.
    • Screenshots of code. Link the repository.
    • Accuracy figures with no baseline to compare against.

Where the page comes from

Start from what you already have.

CV as a PDF

Roles, employers, dates, education, and skills are read from the file. A LinkedIn profile saved to PDF works.

GitHub profile

Public repositories are added as projects. You choose which ones stay on the page.

Courses and certificates

Courses sit on the same timeline as roles, which helps if you moved into data from another field.

The three steps are on the homepage. The price is on the pricing page.

Questions

My work is under NDA. What can I show?

Describe the question and the outcome without the figures, and add one project on public data. You decide what stays on the page, and nothing is published until you confirm it.

I am moving into data from another career. Does the old career show?

Only what you keep. Earlier roles can stay on the timeline, be shortened, or be dropped. Courses and projects appear alongside them.

Can it embed a Tableau or Power BI dashboard?

No. The page links out to the dashboard. It does not embed it.

What does it cost?

Publishing is free, with a small “Made with pathfol.com” bar on the page. One payment of $29 removes it. There is no monthly charge.

Bring a CV. Leave with a link.

The draft is private until you say it is right. Publishing is free on pathfol.com.

Create your profile

Also for: web developers, designers, marketers, product managers.