# How to Find Your First Open-Source Project in One Evening > Forty days before I wrote this, I had never opened a pull request. This is the method I used to choose a repository, and an honest account of what the first contribution feels like. - Author: Yadunandan Singh (https://yadunandan-singh.pages.dev/) - Published: 2026-07-24 ยท updated 2026-10-19 - URL: https://yadunandan-singh.pages.dev/blog/find-your-first-open-source-project/ - Topics: Open Source, GitHub, Career - First published on Medium: https://medium.com/@yadunandan-ai-dev/how-to-find-your-first-open-source-project-in-one-evening-ab936a19d9dd ## Summary Yadunandan Singh explains the method he used to go from never having opened a pull request to eight merged pull requests in the c2siorg organisation (DataLoom and TensorMap) within forty days. Answer three questions on paper (your strongest language, your domain and your realistic weekly hours), turn them into one targeted AI search prompt, then spend 90 seconds vetting each suggested repository. The second half is an honest, step-by-step account of a first pull request, from local setup to review and merge. ## Key takeaways - Decide your language, domain and realistic weekly hours on paper before you search GitHub. - Turn those answers into one specific search prompt: at least 50 stars, your stack, actively accepting contributors. - Vet each repository in 90 seconds: a commit in the last 30 days, reviews within a week, a real CONTRIBUTING.md and a welcoming tone in the issues. - Start with small fixes; showing up consistently matters more than skill on day one. - Review delays are normal: never wait idle, and ping politely once after a week. --- I'm 18. Forty days before I wrote the first version of this post, I had never opened a pull request. By the time I published it, I had eight merged across two projects in the [c2siorg](https://github.com/c2siorg) organisation. None of that came from being unusually talented. It came from a method for the part nobody teaches: choosing *where* to contribute. This post is that method, followed by an honest, step-by-step account of what a first pull request actually feels like. ## The one line that sent me to GitHub at midnight I was half-watching a YouTube video when someone said that open source can completely change how you think about software engineering. I don't remember the channel or the video. I remember the line. I didn't announce anything. No "Day 1 of my open-source journey" post, no telling friends. I just wanted to find out whether it was true. There was another reason, and I'd rather be honest about it because everyone has it: contributing to open source carries status. Every class has one person who works on real projects, and once people find out, it's all they talk about. That's a nice side effect. It isn't a reason. A reason has to survive week three, when you're reading 4,000 lines of someone else's code at 1 a.m. and none of it makes sense. Curiosity survives that. Status doesn't. ## The hard part isn't contributing. It's choosing. Every tutorial says "start contributing to open source." Almost none of them say where. So you search GitHub for "good first issue" and run straight into the real beginner problem: hundreds of millions of repositories, issues that were claimed eight months ago and never closed, projects whose last commit was in 2021, and maintainers who never reply. You bounce between five repositories in one evening and contribute to none of them. That's exactly where I was. Here's what got me out. ## Step 1: Answer three questions on paper Before you open GitHub, get a notebook and a pen and write the answers down. Typing them isn't the same, and you'll need them in Step 2. 1. **Which language am I genuinely good at?** Not learning. Not "did a course in." Good at. If you can't debug in it without looking up the syntax, it doesn't go on the list. 2. **If I had to build one polished project from scratch, what would it be, and in what stack?** This question quietly tells you your domain. If your answer involves dataframes and model training, your repository lives in the ML-tooling world. If it involves routes and authentication, it lives somewhere else entirely. Write the frameworks and libraries down by name. 3. **How many hours a week can I realistically give this, for three months rather than one excited weekend?** Be brutal. Six hours you actually keep beats forty hours that disappear after ten days. Maintainers notice who comes back. Consistency is the real currency of open source, and it's the one thing you can offer on day one that has nothing to do with skill. ## Step 2: Turn your answers into one search prompt Now open an AI assistant (Copilot, ChatGPT and Claude all work the same here) and paste this in, filling the brackets with what you just wrote: ```prompt I'm looking for an open-source organisation to contribute to. Find projects that meet all of these criteria: at least 50 stars, well regarded in [your field], built in [your language], using frameworks or libraries like [your frameworks], and actively accepting new contributors. For each one, tell me the last commit date and how quickly recent pull requests were reviewed. ``` This is almost exactly what I typed: ```prompt I'm looking for an open-source organisation to contribute to. Find projects that meet all of these criteria: at least 50 stars, well regarded in the data/ML space, built in Python or Rust, using frameworks or libraries like pandas, TensorFlow or scikit-learn, and actively accepting new contributors. ``` Without the three answers from Step 1, that prompt is empty. That's why the notebook comes first. ## Step 3: Spend 90 seconds verifying each suggestion Don't trust the list blindly. The model will hand you ten repositories and some of them will be dead. Open each one and check four things: - **Last commit:** within the past 30 days. Any older and you'll wait forever for a review. - **Open pull requests:** are recent ones getting comments within a week? Open two and read the review threads. - **A `CONTRIBUTING.md` that actually exists** and explains how to run the project locally. - **The tone in the issues:** read three issue threads. Within a minute you'll know whether beginners get helped or ignored. That last check matters more than the star count, and almost nobody tells you so. ## Why I picked DataLoom I ended up at [DataLoom](https://github.com/c2siorg/DataLoom), a data-cleaning and transformation tool in the c2siorg organisation. It passed all four checks, it sat exactly in the domain I'd written on paper, and it was small enough that I could eventually hold most of it in my head. For a first project, that matters far more than prestige. My first contribution went in on 15 June 2026. Forty days later I had eight merged pull requests across DataLoom and its sibling project, [TensorMap](https://github.com/c2siorg/tensormap). ![My closed pull requests on the c2siorg/DataLoom repository, including fixes for case-sensitive string filtering, dark-mode styling and the lint workflow, and features such as toolbar tooltips and an apply-preview workflow.](https://yadunandan-singh.pages.dev/blog/find-your-first-open-source-project/pull-requests.webp "My pull requests on DataLoom after the first forty days: small fixes first, bigger features later, and a couple that needed a second attempt.") None of it was glamorous at first. A case-sensitivity bug in string filtering. Hover tooltips on toolbar icons. A dark-mode panel that was still using hard-coded light-theme classes. Then bigger things: a preview-before-apply flow rolled out across all twelve transformation forms, and a CI change that split the lint workflow so untrusted pull-request code no longer runs with write permissions. You don't start at the CI security fix. You start at the tooltip. That's the whole point. ## What your first pull request actually feels like This is the part I wish someone had walked me through, one moment at a time. **You open the repository and feel like the least qualified person alive.** Everyone else's pull requests look surgical. The issue titles are full of acronyms you've never seen. That feeling is not a signal. Everyone on the contributors list felt it, and most of them still do on a new codebase. **You fork it, clone it, and try to run it locally.** This alone can eat your first evening. Mine involved Docker Compose, a Postgres container and a missing environment variable that took an hour to find. When it finally boots, that's your first real win, and it's bigger than it feels. **You find a bug.** A small one, but a real one. You open an issue and write, "Can I work on this?" **Someone replies: "Go ahead, it's yours."** This is the moment everything turns. A stranger with commit rights on a project you admire has just handed you a piece of it, in public, with your username attached. That isn't encouragement. It's permission, and it feels heavier. **You write the fix in forty minutes and spend three hours on the commit message.** Then you press "Create pull request" with a million thoughts in your head, most of them about how it'll look when they reject it. **Then: nothing.** Hours. Sometimes days. This is the stage that breaks people, so read the next section before it happens to you. **Then a comment: "needs revision."** You'll read it as rejection. It isn't. A maintainer spent part of their evening reading code from someone they've never met, and cared enough to tell you how to make it better. It's the most under-appreciated gift in the whole ecosystem. You push a fix in ten minutes and start waiting again. **Then the icon turns purple. Merged.** Code you wrote on your laptop, in your room, while doubting every line, is now running on the machines of people who will never know your name. I won't try to name that feeling. Go and get it. ## The waiting is normal Maintainers are volunteers with jobs, degrees, families and a review queue. A few hours to a few days is completely normal, and silence is almost never about you. Two rules fixed this for me: 1. **Never wait idle.** The moment a pull request goes up, start on the next issue. If you're always a few issues deep, review latency stops being emotionally expensive. 2. **Ping once, politely, after a week.** "Hi, just checking whether this needs anything from my side." That's the whole message. ## Nobody judged me The biggest fear every new contributor carries is looking stupid in public, and it's the permanent, searchable, screenshot-able kind of public. Here's what actually happened. I asked beginner questions and got real answers. I proposed a feature, another contributor pushed back with a better-reasoned design, and I changed direction, without a trace of condescension anywhere in the thread. Before the forty days were up, I reviewed someone else's pull request for the first time and raised two issues with the approach. They fixed one immediately and recorded the other as a follow-up. That isn't luck. It's what a healthy project looks like, and it's exactly what you're screening for when you read those issue threads in Step 3. The person who wrote "go ahead" on my first issue and the maintainer who merged my first pull request have no idea they did anything. They did everything. ## So, has open source changed my life? Honestly, I don't know yet, and I'm not going to pretend otherwise. But here's what I can measure: - I can read a large codebase I didn't write and find my way around it. - I can take a review comment without my ego getting involved. - I've been wrong in public, in writing, in front of better engineers, and nothing bad happened. - I have a public record of real work, which at 18 and without a computer-science degree is worth more than another certificate. That isn't "life-changing" yet. It's a lot more than I had in May. ## Your turn Do Step 1 tonight: three questions, on paper. By the end of the evening you'll have a repository, and not long after, a first pull request. If this is the post that gets you there, I'd genuinely like to hear about it: which repository you picked, and what that first "merged" notification felt like. [Send me an email](mailto:yadunandansingh027@gmail.com) or message me on [LinkedIn](https://www.linkedin.com/in/yadunandan-singh-ai-dev/).