A Languages
- Python
- Java
- JavaScript
- TypeScript
- SQL
- HTML / CSS
CS Undergrad | Developer | Debater | Teacher
I build practical software for real problems: sentiment analytics, mobile finance tools and AI utilities. The rest of my time goes into teaching. Years of debate and public speaking taught me that clarity matters as much in an interface as it does in an argument.
Two halves of the same habit: taking something complicated and making it legible.
class SulavPaudel {
constructor() {
this.role = "CS Undergraduate";
this.focus = ["NLP", "Full-stack", "Mobile"];
this.principle = "Ship something people can actually use";
}
build() {
return this.focus.map(area => solveRealProblem(area));
}
grow() {
return "Take the hard brief. Learn in public.";
}
}
I like problems that survive contact with real users: data that is messy, APIs that rate-limit, models that need tuning rather than a demo that only works once.
Most of what I build ends up somewhere between data and interface, scraping and scoring thousands of comments, then making the result readable at a glance.
“Strong communication is not decoration on top of good work. It is the part that makes the work usable by anyone else.”
My route into tech ran through communication. As a public speaker and debate trainer I learned to value clarity, structure and impact above cleverness.
Whether I am mentoring, chairing a discussion or presenting an idea, the method is the same: listen properly, structure the argument, then say it plainly.
That carries directly into engineering: naming things well, writing docs people read, and explaining trade-offs without hiding behind jargon.
Teaching, engineering and operations, filtered by kind of work.
What I reach for, grouped by where it sits in the stack.
Selected work: built, shipped and still standing.
Web App
A Flask app that measures how the public feels about any topic. Gemini picks the relevant subreddits and search queries, then background workers mine roughly 90 days of Reddit and YouTube comments behind a live progress bar.
A fine-tuned RoBERTa model scores every comment. Results are stored in Supabase, aggregated into one PSI score and cached so dashboards load instantly. On top of the numbers sit AI deep-dives, an “Ask AI” endpoint grounded in the real comments, and a versus mode that overlays two topics on one timeline.
Mobile App
A privacy-first personal finance tracker. Log and analyse spending across cash, bank, eSewa and Khalti in one place.
Multi-source tracking with graphical reports, and every record stays offline on the device, with no account, no sync and no server.
Extension
A Chrome extension that turns what you just watched into a quiz. It detects the watched portion of a video and generates contextual questions from captions and timestamps.
Published on the Chrome Web Store and powered by Gemini 2.5 Flash behind a FastAPI backend.
Web App
A modern matchmaking web app built as a college project: traditional matchmaking logic wrapped in a swipe-based interface.
Clean frontend over a structured Java backend with a REST layer and relational schema.
Hiring, collaborating or just arguing about something. All welcome.