A thought on why your portfolio should be an expression of thought - not a one-time performance.

I bought my first domain - PrabakaranChandran.com - sometime around 2019 or 2020. I wasn't great at frontend or backend back then, so I just built things with WordPress. And honestly, it worked. I had an overview of my projects from Mu Sigma, some blogs introducing basic concepts, curated reading lists - because I always believed if somebody has already written something well, just share it with people, right? What's funny is, even now, six or seven years later, I look back at those posts and think: did I really write all that? Some of it still holds up. It's not obsolete. And that alone tells you something about what a portfolio can do when you let it live long enough.

Gateway, Not a Page

So here's the thing. I wouldn't even call it a "portfolio website." That phrase carries a very specific, very limited image - one page, some projects, a resume, done. A showcase piece. But I really believe a portfolio is a gateway. A gateway to understand a particular person - what they think, what they build, what they care about, where they're heading. And the way I've arrived at this belief is by observing how very different kinds of people maintain their portfolios over the years, and what that reveals about the discipline and intention behind it.

If you look at the very software-nerdy, coding-focused builders - the people who always say "we should ship things" - for them, sometimes a GitHub profile alone serves as a great portfolio. They ship, the work speaks. But the moment you want to bring some identity beyond code - what you read, what you think, what interests you outside of engineering - you need something more. So people wrap it all into one website: books they're reading, their opinions, their side interests, the whole thing. And that's when the portfolio starts becoming something richer than just a project gallery.

Now, if you look at the other end - researchers, professors, PhD students - this is where I've seen the most structured and consistent approach. If you check any Columbia ML or CS professor's website, you'll find publications, teaching, talks, and a very curated set of resources they believe will help students. Things like "how to write a good research paper" or "different perspectives on machine learning." I really like going through some of my favorite professors' websites, because you always find something structured, something that gives you good resources to read, something that reflects their particular point of view. Even PhD students follow a very consistent baseline - list of publications, tutorials or talks they've given, blogs they've written, projects they've built. Almost always on GitHub Pages, often with Jekyll. It's clean, it compounds, and it tells you exactly where that person stands intellectually.

The typical software engineers, though - and I say this with love, because I've been here - tend to have scattered portfolios. Some stuff on GitHub, some on LinkedIn, maybe a personal site that hasn't been touched in two years. The discipline of keeping everything in one specific place just doesn't come naturally when you're deep in client work or shipping production code. My own GitHub is chaos, if I'm being honest. Very sparse. I never built that discipline early on. But what I did maintain was writing on LinkedIn and keeping my personal website alive. LinkedIn, despite all the performative noise and the dilution that's happened over time - it's still a very good ecosystem. A lot of people exist in one place, and you can share your journey transparently. Not just "I got a job" or "I went to a hackathon," but the actual arc of what you've been thinking and learning. I still believe LinkedIn is a very good showcase of what I've been doing, because it's more or less a transparent journey. And I know people might have very different opinions or even hate on LinkedIn, but I don't have any sort of bad feeling about it. It has its problems, sure, but the ecosystem is real.

Don't Build a Portfolio. Grow One.

Now, here's where I have a strong opinion, and this is really the core of what I want to say.

A portfolio should not be built for job seeking. It should not be a project you complete in a week. I've seen this pattern too many times - someone spends three or four weeks, builds some fancy JS animations, puts together two or three replicated papers, maybe "Attention Is All You Need" or some random paper implementation, drops it all on a website, and calls it done. Maybe they announce it on LinkedIn: "I'm glad to share my portfolio!" And then? Nothing. It doesn't evolve. No new content, no new ideas, no new writing. It just sits there like a frozen artifact. The ingredients were assembled - some projects, some website - and the portfolio was declared "ready." But that's treating it like a deliverable, and I think that's fundamentally the wrong frame.

Building projects just for the sake of putting them on a resume - I don't think that's great advice. I mean, people will disagree, and that's fine, it's something to be discussed. But replicating someone else's paper implementation just to have a GitHub link on your portfolio? That's not really your thought. That's not your expression. What I'd suggest instead is to build what you have in your mind. If you've been thinking differently about reasoning, or inference, or some machine learning algorithm - build that. Even if it's half-formed. Even if it takes months to get more understanding, more validation on the concept you've been learning. That's the real stuff. The self-organizing, slowly crystallizing kind of work. Over the months, you get more clarity, some validations, and it becomes genuinely yours. Not a copy-paste, not a replication, but something that emerged from your own thinking.

Expression of Thought

And this connects to something deeper that I keep coming back to. There's a quote from Jiddu Krishnamurti - we fondly call him JK - where he says that everything is an expression of thought. A computer is an expression of thought - thought from Babbage, from Turing. That expression keeps getting refined, others adopt it, get inspired, and they build further. The expression of thought keeps evolving. And I think a portfolio should be exactly that - a very good expression of your thought, and a system of execution.

Think about it. Your thoughts, your projects, whatever you learn, whatever you read. Maybe you have a difference of opinion on some concept. Maybe you've dissected a paper and found something interesting. Maybe you have a new idea that's still developing - what I call a draft of thought. Even your traces of thinking, your work in progress, your half-baked explorations. All of this can live in one place. You can call it portfolio, you can call it whatever you want. The point is that it's cumulative. It compounds. And it gives you something you can't get any other way - the ability to look back and say, oh, this is what I've really worked on, this is where I went wrong, this is where I am, this is where I'm heading.

Look at Paul Graham's website. Look at Sam Altman's. Dead simple. Very static. But the content is incredible - years and years of accumulated thinking. Or take Professor Andrew Gelman, the famous statistician and political scientist at Columbia. He's been writing on his blog for years. He keeps sharing his ideas, his observations, his opinions on various papers and concepts, and people refer to it, conversations happen around it. That blog is his portfolio in a deep sense. It's not a showcase piece - it's a living, breathing expression of how he thinks. And his expression of thought extends even further - the QMSS department at Columbia, his books on Bayesian data analysis - it's all connected, all flowing from the same intellectual system.

I'm not saying everyone should build something that lively or dynamic, where a lot of people come and visit and engage. No, it can be very personal. But at the very least, it has to be long-living and continuously operating - not something where you build it, put it up, share the link when people ask, and forget about it.

Three Ways It Goes Wrong

So let me put it plainly. There are three ways a portfolio goes wrong, and I've seen all three enough times to feel confident naming them.

The first is what I'd call the showcase piece. Built once with flashy animations and all the right visual elements, but never updated. It looks impressive for a week, then it becomes a digital fossil. You're not adding content, not sharing knowledge, not evolving. It's a one-time performance, and performance fades.

The second is the resume annex. This is when projects exist only because they needed to exist on a resume. No real curiosity behind them, no personal stake. The portfolio becomes a checklist - "oh, okay, I need two projects, a website, and a deployed link, then I'm done." But that's treating the portfolio as a project with a deadline, and portfolios shouldn't have deadlines.

The third is the platform substitute. This is when you rely entirely on LinkedIn or Twitter or GitHub as your portfolio. These are great channels, genuinely useful, and I use them myself. But they're someone else's platform. You don't fully control the structure, the narrative, the longevity. They should complement your portfolio, not replace it.

System, Not Substance

What I'm really advocating for is something different from all three of these. Think of your portfolio as a system - not a substance. A system that helps you keep things structured. A system that keeps your thinking wide, transparent, and evolving. A system where you write notes, share blogs, express ideas, and over time, it gives you a better and better understanding of your own trajectory. It's self-provisioning - nobody's going to come and comment and guide you through it. Even if nobody visits, even if no one reads it, the act of keeping things in record gives you a quiet kind of guidance. It compounds in ways you can't predict.

I've made plenty of mistakes here myself, and I want to be honest about that. I built websites and didn't maintain them. I explored different identity containers - beyond.fit, nth experiment, we cook labs - trying to express ideas under different names, different framings. But eventually, I convinced myself that what I needed was just one consistent system. Keep building it, keep growing it. That's why pracha.me exists now. Prabakaran Chandran, reduced to Pracha. Simple. One place where daily.pracha.me carries the daily writing, where the subdomains carry the different expressions, but it's all one system.

The Practical Bit

And if you're actually going to set something up - the practical bit - my honest suggestion is to just use GitHub Pages. Seriously. For a portfolio, all you really need is static HTML, maybe some JS to manage a few things. You can use Jekyll templates, which give you a very structured way to manage blogs, projects, and publications. Write in Markdown, commit, and it publishes automatically. Set up a GitHub Actions workflow so that whenever you add a new markdown file or some details, it goes live. Find a domain - $5 to $10 in the US, maybe Rs. 500 in India. Keep it `.me`, `.io`, `.com`, or even just `.github.io` if you want zero cost. If you need something running in the backend, some compute, you can look at GCP Cloud Run - I used to host on Cloud Run and it's very cheap. But for most people, static GitHub Pages is more than enough.

Be as curious, as expressive, as creative as you want with how you structure it. But just keep it alive. Think of it as a system, not a one-time performative substance.

Keep It Alive

Because here's the thing - a portfolio is not a deliverable. It's not something you build in a sprint and ship. It's an evolving expression of who you are - what you think, what you build, what you question, what you learn. It gives you an evolving personality, because you've been expressing and working and reflecting through it. And over time - months, years - it becomes something you couldn't have planned. Something that's genuinely yours.

So yeah. Keep it simple. Keep it honest. Keep it alive.

That's the whole point, right?

References