When I received the news that our team, The Dobermans, had been selected for the eSewa Hackathon 2026, I was excited beyond words.
More than 150 teams had applied, and only a small number were selected to compete at the national level. We chose Challenge 4: Smart KYC Assistant.
But beneath that excitement was something I didn't tell many people.
I was scared.
Not because I couldn't code.
I knew how to build applications. I knew how to turn ideas into working products. I knew how to create MVPs quickly and solve problems under pressure.
What I wasn't confident about was Machine Learning.
The challenge required much more than building screens and APIs. It required understanding document verification, fraud detection, image processing, identity validation, and AI-assisted onboarding.
Every time I thought about those areas, a voice in my head kept asking:
"What if I can't do it?"
"What if my skills aren't enough?"
"What if I get exposed?"
For a moment, part of me wanted an excuse to avoid the challenge altogether.
Thankfully, I wasn't alone.
My teammates pushed me forward.
One of my closest friends shares the same obsession that I do: the pursuit of mastery. The desire to reach the highest level possible.
Whenever doubt appeared, They reminded me why we were there.
Not to avoid difficult problems.
To solve them.
The Weight of Previous Failures
This wasn't my first hackathon.
In previous competitions, I hadn't contributed the way I wanted to.
I relied too much on others.
I watched opportunities pass by because I wasn't taking enough ownership.
And honestly, I hated that version of myself.
The eSewa Hackathon became my opportunity to change that.
I wanted proof that I could carry responsibility.
Proof that I could build.
Proof that I wasn't just someone who talked about ideas but someone who could execute them.
Over the course of the hackathon, I learned something important:
I can code.
I can build under pressure.
I can ship products quickly.
But most importantly, I can do those things professionally.
An Industry-Level Experience
The hackathon itself was unlike anything I had experienced before.
Held in Sauraha, Chitwan, the event brought together talented developers, designers, and innovators from across Nepal.
The atmosphere felt different from college competitions.
This felt like the real industry.
The venue, mentors, judges, discussions, and expectations reflected professional standards.
What impressed me most was how we were treated.
Not as students.
Not as beginners.
But as developers.
People genuinely wanted to hear our ideas.
They challenged our assumptions.
They questioned our architecture.
They pushed us to think beyond prototypes and toward production-ready solutions.
That level of respect was incredibly motivating.
For the first time, I felt what it might be like to work in the technology industry at a serious level.
And honestly, it made me want more.
Much more.
Building Smart KYC Assistant
The challenge we selected was Smart KYC Assistant, an AI-powered identity verification platform designed for Nepali fintech applications.
The idea started with a simple question:
Why are users still manually typing information that already exists on their citizenship card?
Every fintech platform requires KYC.
Most users spend several minutes entering information that already exists on their documents.
Mistakes happen constantly.
Rejected applications increase.
Users abandon onboarding.
We wanted to eliminate that friction entirely.
The solution we built follows an eight-step onboarding flow:
- Upload identity document.
- EasyOCR + Tesseract reads both sides of the document.
- AI extracts every KYC field automatically.
- Users review extracted data beside the original document.
- Live face verification through webcam.
- AI validates identity and liveness.
- Real-time KYC status tracking.
- GPT-4o assistant provides support in Nepali and English.
Instead of manually typing information, users simply upload their citizenship card.
Within seconds, the platform extracts names, citizenship numbers, dates of birth, addresses, provinces, districts, municipalities, wards, and additional identity information.
A review screen displays the uploaded document side-by-side with extracted data, allowing users to verify everything before submission.
After confirmation, the user completes live face verification.
The platform performs identity validation, liveness checks, image-quality assessment, and fraud-prevention verification before continuing.
Users can then track their KYC application in real time.
Instead of receiving vague rejection messages, they receive detailed AI-generated explanations and corrective guidance.
We also built a GPT-4o-powered assistant capable of understanding the user's actual KYC status and responding in both Nepali and English.
The technical stack:
- React - frontend interface and onboarding flow
- FastAPI - backend API layer
- PostgreSQL - persistent storage
- GPT-4o - conversational assistant and status explanation
- EasyOCR + Tesseract - document text extraction
Session state persists across browser refreshes, ensuring users never lose progress during onboarding.
By the end of the hackathon, we were no longer thinking about the project as a prototype.
We were thinking about it as infrastructure.
A system that any fintech company in Nepal could integrate into its onboarding workflow.
What I Took Away
The biggest lesson from eSewa Hackathon wasn't technical.
It was personal.
Fear disappears when you move toward it.
The things that look impossible from a distance become manageable once you start working on them.
Before the hackathon, I was worried about machine learning.
After the hackathon, I wasn't suddenly an ML expert.
But I was no longer afraid of it.
And that is progress.
Today, when I look back at the experience, I remember growth more than stress.
I remember responsibility.
I remember execution.
Most importantly, I remember proving something to myself.
The eSewa Hackathon didn't just help me build a project.
It helped me build confidence.
And for that reason alone, it remains one of the most important milestones in my engineering journey.