Portfolio Examples · Data Science Portfolio

Data Scientist Portfolio Example and Template: A Sample Applied ML Portfolio for 2026

Updated November 3, 2026 · 9 min read · By Seera

This page is a fully built sample data science portfolio website — built on the same Pulse template a paying Seera client would publish. The layout, animations, and styling are byte-for-byte identical to what you would get if you published a data science portfolio on Seera tonight. The only difference is the data: instead of a real practitioner's projects, this example uses a fictional but realistic profile (Ravi Krishnan, a Berlin-based senior data scientist working on applied ML in healthcare with prior stops in fintech and e-commerce).

Short answer: a data scientist portfolio needs three to five project case studies, every one on the same card (the problem, the training data, the model and how it was evaluated, the result, what happened after launch, and links), with the code on GitHub and a website that ties it together. A PDF is only for portals that ask for a file. Want real portfolios instead of a sample? Read seven real data science portfolio examples that we opened and reviewed one at a time, or open the live portfolios published on Seera.

What you can take from this data science portfolio example: the section structure, how to present a production-deployed model alongside open-source work, what depth a fraud-detection or recommendation case study should have, and how a senior IC differs in framing from a research-leaning data scientist.

View the Live Data Science Portfolio →

The data science portfolio at a glance

Data ScientistRavi Krishnan (fictional, modeled on a working pattern)
SpecialtySenior IC · applied ML · production deployment
IndustryHealthcare, fintech, e-commerce
LocationBerlin, Germany — open to remote (EU/UK)
Years working8 years
Portfolio templatePulse — animated cyan-violet on deep navy
Sections shownHero · About · Stack · 4 Project Case Studies · Experience · Education · Testimonials · Contact

Why this layout works for a data science portfolio

The Pulse template paired with a cyan-violet palette is a deliberate choice for an applied data science portfolio. Three things about why it works:

Section-by-section breakdown of the data science portfolio

1. Hero — role, focus, and current availability

The hero reads "Senior Data Scientist · Applied ML & Production Systems" — specific enough that an engineering manager hiring for an applied-ML role knows immediately whether to keep reading.

2. About — operating style, not biography

The about section explains the operating style: production-and-applied, selective project list, and quarterly deep-dive technical writing. Crucially, it shows current availability above the fold.

3. Tools and methods — grouped by use case

For any data science portfolio, this section answers two practical questions: do they ship in the languages and tools we use, and do they understand the production layer underneath. The example breaks the stack into Specialties, Languages & Modelling, Production & MLOps, and Practice. The Production & MLOps grouping is what separates a senior IC from a research-only data scientist.

4. Project case studies — depth over breadth

The case-study section is the strongest part of any data science portfolio. The example here shows four projects across deliberately different categories:

A data science portfolio of four Kaggle competitions shows you can do Kaggle competitions. A portfolio of four different kinds of work — production-deployed, A/B-tested in production, open-source, and applied recommender — shows a data scientist with range that maps to senior and staff roles.

5. Work history — companies, roles, specific contributions

Each role on this data science portfolio leads with the most concrete thing the practitioner did there: models shipped, false positives reduced, AUC achieved, juniors mentored.

6. Testimonials — from CMOs, heads of risk, and open-source maintainers

Three testimonials in this data science portfolio: a CMO (the healthcare client), a former Head of Risk (the fintech client), and an open-source maintainer (independent verification of the library work).

7. Contact — one email, one clear next step

The contact section is intentionally simple: an email, GitHub, LinkedIn, and a clear note on availability.

What this data science portfolio gets right (and what to copy)

  1. Production-deployed work, leading. The clinical triage model — deployed in 12 hospitals, monitored, audited — is the strongest possible signal.
  2. Open-source work treated as headline, not side note. The uplift library has its own case study with stars, benchmarks, and adoption.
  3. Experimental rigour stated explicitly. "Pre-registered A/B test in production over 60 days" — this is what separates a working data scientist from a notebook tourist.
  4. Bias audit mentioned upfront. The healthcare project explicitly mentions a bias audit by ethnicity, age, and gender — credibility-building for sensitive deployments.
  5. Range of categories. Healthcare, fintech, e-commerce, open source — signals what the data scientist can be hired for.

Data science portfolio website template: what it needs

A data science portfolio template needs a one-line hero, project cards that give every project the same fields, tools grouped by use, links to the code and a demo, and one way to get in touch. Identical fields are what make five projects read as one body of work. Copy these six for each project card:

Project card field What it holds
ProblemThe question or decision the model supports, in one sentence
Training dataSource, size, date range and licence
Model and experiment designThe approach, the baseline it beat and how it was evaluated
ResultThe metric, with its scope and the period it covers
Monitoring or limitsWhat happens after launch, or what the result cannot tell you
LinksCode, a demo or a write-up a stranger can open

Is there a free data science portfolio website template? Every template can be opened and previewed for free, and the one that also goes live without paying is Sovereign. Pulse, the layout used in this example, is a Pro template, billed at €4.99 a month yearly.

Data science portfolio on GitHub, GitHub Pages or a website?

Use GitHub for the code and a website for the summary. A profile shows what you built but not why, and GitHub lets you pin up to six repositories and gists combined. GitHub can also publish a static site from a repository, at owner.github.io, which is how some portfolios are hosted. Hosting is not the dividing line: in our review of seven data science portfolios, four are hosted on GitHub Pages, and the strong ones add a plain-language title, a sentence on the problem and a clickable link to each project.

Data science portfolio PDF: when to use one

Use a PDF when an application portal asks for an attachment. It is a snapshot: it cannot show a live demo and goes stale as soon as you send it, so make it a short summary of your best three projects with working links, and keep the website as the source. On Seera, PDF export is part of Pro.

How to build a data science portfolio like this for yourself

  1. Upload your CV or data scientist bio to Seera. The AI extracts your projects, models, and stack.
  2. Pick the Pulse template (or browse the other 23 templates — DevTerminal for research-heavy, Glass for industrial, Aurora for premium analyst).
  3. Replace the four sample case studies with three to five of your own — at minimum one production-deployed model, one open-source piece, one A/B-tested project.
  4. Write proper technical case studies. Each project needs model architecture, training data, experiment design, result, and post-launch monitoring or limitations.
  5. Get three named testimonials — a manager, a peer or open-source contributor, and a downstream stakeholder (clinician, risk manager, PM).
  6. Connect your custom domain on Pro. ravikrishnan.com signals seriousness in a way a builder subdomain never will.
Build Your Data Science Portfolio Free →

Frequently asked questions about data science portfolios

What does a good data science portfolio look like in 2026?

A strong data science portfolio in 2026 leads with applied projects that shipped to production — three to five case studies with the model, the experiment design, the deployment architecture, and the post-launch monitoring — rather than Kaggle leaderboard placements. The example here is a senior data scientist's portfolio built on a clean, animated, data-vibe template (cyan and violet accents on deep navy). Avoid: long lists of competitions entered, ten Jupyter notebooks without deployment context, and 'I love data' opening paragraphs.

What sections should a data science portfolio include?

Six core sections: (1) Hero — name, role, focus, and current availability; (2) About — short positioning paragraph on the kind of data science work; (3) Project case studies — three to five applied ML projects with measurable outcomes and production architecture; (4) Tools and methods — modelling stack, MLOps stack, and production tools grouped by use case; (5) Work history — companies, roles, and the specific contribution at each; (6) Testimonials and education.

What projects should a data scientist include on a portfolio?

For a senior or staff data science portfolio, three to five applied projects with depth beats a Kaggle leaderboard list. Aim for one production-deployed model (the highest signal — verifiable, monitored, used), one open-source contribution (independent verifiability), one A/B-tested project (experimental discipline), and one classic applied piece (a recommender, fraud detector, or churn model). Junior data scientists should lead with two end-to-end projects deployed somewhere — even just a Streamlit demo with a writeup.

Which template style works best for a data science portfolio?

For senior data scientists working on applied ML, the Pulse template (animated, cyan-violet, data-vibe) is a strong fit — it signals technical seriousness without going full developer-terminal. The example portfolio on this page uses Pulse in a cyan-and-violet palette. Data scientists doing more research-heavy or paper-led work may prefer DevTerminal (terminal aesthetic, monospace) or Glass (dark glassmorphism). Data analysts working on dashboards and reporting may prefer Aurora (premium glass) or Gradient (modern data-friendly).

How do I write a data science portfolio bio?

A data science portfolio bio should answer four questions in two short paragraphs: (1) what you do — applied ML, recommendation systems, NLP, causal inference, ML engineering; (2) the kind of company you work at — e-commerce, fintech, healthcare, research; (3) the operating style — production-and-applied, research-first, end-to-end; (4) what is currently true — open to senior roles, on a contract, available from a date. Avoid: 'passionate about data', 'data wizard', 'machine learning enthusiast'.

How important is open-source work for a data science portfolio?

For senior and staff data scientists, open-source work is one of the highest-leverage signals on the portfolio — it is independently verifiable, it survives company-confidentiality issues, and it shows how you collaborate with strangers. The example portfolio on this page leads with an open-source uplift-modelling library because the README, the benchmarks, and the maintainer-review history are all visible to any hiring manager who clicks through. Even one well-documented library outweighs a longer list of company projects you cannot talk about.

What should a data science portfolio website template include?

A one-line hero, project cards with the same six fields (problem, training data, model and experiment design, result, monitoring or limits, links), tools grouped by use, and one way to get in touch. Identical fields across projects make the portfolio read as one body of work.

Is GitHub enough for a data science portfolio?

A GitHub profile proves you can code and shows little else. Keep the code on GitHub, pin up to six repositories, and put a short summary of each project, with the problem, the result and a link, on a portfolio page. Four of the seven portfolios reviewed in our data science portfolio examples are hosted on GitHub Pages, so framing matters more than hosting.

Should a data science portfolio be a PDF?

Only as an attachment. A PDF cannot show a live demo and goes stale once sent, so use it when an application portal asks for a file, keep it to a short summary of your best three projects with working links, and make the website the source.

Related data science portfolio resources