Why building an Award-Winning Project not as hard as it seems

Hi, I’m Aditya, a high school junior from Mumbai, India. My journey in STEM started in 8th grade with a robotics project that I started solely because I was inspired by Cisco Ramon after watching all seasons of The Flash during the COVID-19 lockdown. I went on to win the BROADCOM Coding with Commitment challenge at my national fair in 2022. I have competed at India’s top research fair, IRIS for four years now, winning the grand award thrice.

Winning at ISEF 2023 in 9th grade was a turning point in my high school journey, followed by a series of other national and international competitions I competed in over the next few years. This story from last fall shows that anyone can make a winning science project—if they’re willing to put in the work, no matter their resources. I’m returning to ISEF for the second time this year in Columbus to compete in Computational Biology.

Order and Chaos

Let's go back to the middle of my junior year. I had just moved to Mumbai, a city I barely knew, and was trying to adjust to a new school and environment. On top of that, I had just returned from a month-long research program in August, only to face midterms in September. If that wasn't enough; the submission date for the IRIS National Fair— the ISEF-qualifying fair in India— was at the end of October. To give you a small context about how competitive IRIS is; they receive 2000-3000 preliminary online submissions to compete for 100 project spots to present in one of the 21 subject categories at the national fair every year. Out of all those submissions, just 20 qualify to represent the country at ISEF. That has a success rate of <1%. I had a month to conceptualize, execute, and submit my project. No pressure, right?

I knew I wanted to do something computational. My experience with both lab and computational research has shown me that while lab work has its charm, computational research is faster, more flexible, and often just as impactful. But with just my laptop and internet access, the odds felt stacked against me. What carried me through was something as simple as a basic belief: as long as I put in the hours, to learn new things that come my way, I could make it work.

Murphy's Law

My midterms had finally ended. By mid-October, I had a rough outline of my project: Using histopathological data to detect metastatic signatures in early-stage colon cancer. But a "rough outline" doesn't cut it when the deadline's in two weeks. My baseline model, CatherNet-0, was barely working, and every parameter tweak seemed to push me further away from my goal. A steady companion by then was frustration. However, I found a research paper using a teacher-student model for anomaly detection in tumor patches of breast cancer. I had never heard of this algorithm before but it was pretty promising, so I decided to spend four days replicating it for my colon cancer dataset.

It did not work. At all.

When I re-read the paper, I realized why: their dataset had labelled metastatic patches for training, while mine did not. It was the exact opposite of what I needed. The goal of my study was to show the potential of DL models to predict anomalies in unannotated data. With just five days left, my project was in shambles. My paper? Unwritten. My video submission? A distant thought. My supplemental documents? Not even outlined. Everything that could go wrong, went wrong.

Locked In

From the afternoon of the 26th of October, I worked 32 hours straight, fueled by coffee and sheer determination, and pored over tens of ML papers and algorithm techniques I could use for my project. That’s when I found my saviour–autoencoders, swooping in at the last minute to save the day. In another twelve hours, I validated my hypothesis using secondary data with satisfactory results. The next day, I spent 16 hours writing my paper and performance analysis report, as well as making my video submission.

By some miracle, I submitted everything just in time. Some may question the quality of the research paper or results I submitted, and honestly, I won't argue. Thus, I continued working after my submissions as well to make my model and results better. My goal was not IRIS alone, it was merely an incentive for me to start. When the results came out in mid-November, not only did I make the top 50, but later I also won the grand award in my category, securing my spot to represent India at ISEF 2025. To this day, I’m convinced October was actually three months long—condensed into four frantic days. 

What I Want You To Understand

1. Start with What You Have: Don't let the lack of resources stop you. I did not have some fancy lab or high-end equipment. All I had was a laptop, publicly available datasets, and the internet. That was enough.

2. Learn as You Go: When I found that autoencoder algorithm, I had no idea what any of those words meant. But I didn't hold my horses. I spent all my hours reading, watching video courses on YouTube, experimenting, and troubleshooting. Each step taught me something new.

3. Seek Inspiration Everywhere: Sometimes the breakthrough comes from reading an unrelated paper or technique. Keep your mind open to new ideas because you never know what might click.

4. Time Management: Procrastination may be a fantastic motivator, but it's not a tactic I'd recommend. Still, even when all the time's running out, task prioritizing and doing only what's important can become the difference that saves the day. I had a lot of missing elements during the preliminary submission but what mattered was that I showed the importance of my research through results and my video.

If there’s one thing I want you to take away from my story, it’s this: you don’t need to be a genius to make an impact or do something out of the box. All you need is curiosity, determination, and a willingness to learn. In 2025, the world of research is more accessible than ever, with free resources, open datasets, and online communities ready to support you. Your biggest investment is your time and effort.

If I can do it, you can do it too. And remember, if someone asks how long your project took, feel free to say three months. Some secrets are better kept between you and your laptop:)

Unlock Your Potential with ScienceFair.io!

Ready to elevate your science fair project? ScienceFair.io is your ultimate resource for success, offering a comprehensive platform tailored for high and middle school students. Our tools simplify your project, from brainstorming innovative ideas to presenting your final work with confidence.

Schedule a consultation call to connect with one of our team members and create a personalized plan tailored to your needs. Contact support@sciencefair.io if you have any questions!

What We Offer:

• AI Discovery Engine: Generate unique research ideas based on thousands of successful projects.

• Step-by-Step Research Guide: Navigate the research process with easy-to-follow steps that promote creativity and critical thinking.

• Tuner Features: Optimize your project title and abstract to ensure your work stands out to judges.

• AI Judging Tool: Prepare for the judging experience with simulated Q&A sessions to anticipate and address potential weaknesses.

• Compete Page: Get insights for competing in the International Science & Engineering Fair (ISEF) tailored to your region.

• Expert Coaching: Receive personalized mentorship from past winners through our A-Z Science Fair Masterclass and coaching calls.

Ready to Transform Your Science Fair Journey?

Explore the powerful tools at ScienceFair.io today! Sign up now to unlock your potential and start your successful scientific adventure. Don’t wait—your path to victory starts here!