A Coursera certificate shows that you took and passed a course. That is useful, and there is no reason to hide it. But put yourself in a recruiter’s shoes for a moment: they see dozens of certificates every week, often the same ones. What really catches their attention is being able to see what you can do. A project they can open, browse, try out, and which shows them how you think.
The good news is that you don’t need to invent these projects from scratch. The Atlas paths and curricula already contain plenty. It is mostly a matter of spotting them, making them your own, and presenting them.
Where to find your projects
The first source is capstones. Most specializations end with a final project that requires you to draw on everything you learned in the previous courses. It is the natural starting point for a portfolio.
The second source is the semester projects in the curricula. Each semester of the six Atlas curricula has one, often more personal and open-ended than the capstone. A few examples give an idea of their variety. In Data Science, semester 4, you carry out a complete study in R, from framing the question to a linear model or generalised linear model, including residual checks. In Computer Science, semester 3, you build a mini Unix shell in C, after the recommender system from Duke’s course. In Economics & Finance, semester 5, you build an investment portfolio and backtest it in Python on twenty years of data. In Business & Management, from semester 1, you carry out a diagnosis of a local SME. And in Humanities & Social Sciences, semester 3, you run a complete survey, from the questionnaire to analysing the answers.
Turning an exercise into a personal project
A capstone handed in as is has one flaw: it looks like those of thousands of other learners who took the same course. The recruiter may already have seen it. So you need to give it your own stamp. Three simple approaches do the job.
The first is to change the data. Take the method taught in the course, but apply it to a dataset you really care about: your city, your industry, a hobby. The method stays the same, but the result becomes unique, and you will enjoy talking about it far more.
The second is to go one step further than what was asked. An extra visualization, a second model compared with the first, a small interface that lets people use it. That extra shows you can take the initiative.
The third, perhaps the most effective, is to start from a real question. Rather than an exercise statement, ask yourself something that intrigues you: “have rents risen faster than wages in my city over the past ten years?” A project that answers a real question reads like a story, and that is what makes it memorable.
One important precaution: respect Coursera’s honor code. Don’t publish the solutions to graded assignments as they are, which would amount to handing them to future learners. Publish your own extensions instead.
What each project should contain
Every project deserves a presentation page. On GitHub, a well-written README file is more than enough. That page should answer, in order, a few simple questions: what was the starting question, in one or two sentences? What data did you use, and where does it come from? What approach did you follow, in a few steps? What is the result, illustrated with a chart or a screenshot? What would you do next, with more time? And which tools did you use?
Keep in mind that a recruiter usually spends one or two minutes on a project. The result must therefore be visible straight away, at the top of the page. Technical details can come afterwards, for those who want to dig deeper.
Where to publish
Where you publish depends on the kind of project. For code and data, GitHub (or GitLab) has become the norm: one clean repository per project, with its README. For analyses, published Jupyter notebooks or an online dashboard made with Tableau Public or Looker Studio let you show results interactively. For written work, very common in the humanities, economics or business, a carefully formatted PDF or a blog post will do perfectly.
Finally, create a home page that ties it all together. It can be a very simple personal site or your LinkedIn profile. What matters is that people find, at a glance, three to five selected projects, with a link to each.
How many projects?
Fewer than you think. Three good projects are worth far more than ten exercises. Choose them to cover the skills of the job you are aiming for. For a data analyst role, for example, a data-cleaning project, a SQL analysis and a dashboard make a coherent set that shows the three sides of the job.
Start now
Don’t wait until you have finished your path to publish. Put your first project online as soon as it is done, even if it seems imperfect. You will improve it later, and it is very instructive to see, a few months on, everything you would now do differently. A portfolio that grows month after month is also a great source of motivation, one of the best weapons against dropping out, as we explain in why people drop out of MOOCs.