Yashwi Bhatt

Computer Engineering  ·  LJ University  ·  Ahmedabad
2nd Year  ·  4th Semester

     

I learn best by building things — real things, with real constraints, that break in real ways. My focus is AI / ML, Data Science, and Backend Development: understanding systems deeply enough that the tools become secondary to the thinking.

I care about fundamentals. I care about getting it right. I take the slow path when the slow path is the right one.

→  Strengthening ML foundations — preprocessing, model evaluation, feature engineering
→  DSA grind on LeetCode — consistent, no shortcuts, 75+ problems
→  Deepening backend knowledge — REST APIs, auth, system design basics
→  EDA on real datasets — turning raw data into something that actually says something

Reflecto  ·  Java MySQL JDBC Custom DSA OOP

A personal productivity system — Journal, To-Do, Challenges, Posts, and Auth — built entirely from scratch across 2,470 lines of Java.

 ✔  Data structures implemented by hand: Linked List, Stack, Deque  ✔  Custom exception framework — LengthException, DigitException, EmptyDataException  ✔  MySQL via JDBC with stored procedures for all CRUD operations

Built every part of this without leaning on existing libraries for core logic — because I wanted to understand what was actually happening underneath, not just call a method and move on. Custom exception framework. Clean module boundaries across five independent features.

Every bug was mine to find. Every fix, mine to figure out. The project I'm most proud of, and the most honest one on this profile.

SkillLink  ·  Django Python SQLite OTP Auth Role-Based Access

A multi-role platform connecting Learners, Tutors, and Clients — three separate dashboards, a live-database chatbot, and a complete hiring workflow.

 ✔  OTP email verification and automated booking confirmations  ✔  Live-database chatbot querying real records — not static responses  ✔  Full role-based access control across Learner, Tutor, and Client flows

The most real moment wasn't a feature. It was the night before presenting at Innovation Village: project broken, every fix creating a new problem, an industry panel the next morning. We stripped it down to what mattered and shipped a working version overnight. That night taught me more than any lecture has.

Netflix Content Analysis  ·  Python Pandas Plotly Seaborn EDA

Exploratory analysis on 8,000+ Netflix titles — content trends, regional distribution, and genre patterns over time.

 ✔  8,000+ titles analysed across content type, country, and genre  ✔  Interactive visualisations built with Plotly  ✔  Trend, release pattern, and country-level distribution analysis

Raw data → structured insight. The goal wasn't just to analyse — it was to make the findings actually readable. Every chart was chosen because it answered a specific question, not because it looked good.

Languages  

AI / ML / Data  

Backend  

Databases  

Frontend  

Tools  

Most people think the hardest part of software is writing code.

It's not. The hardest part is understanding the problem clearly enough that the code becomes obvious.

I've learned this the hard way — through 2,470 lines of Java built from the ground up, through a project that broke the night before a demo and had to be rebuilt by morning, through 75+ LeetCode problems where the point was never the number.

I'm not chasing frameworks. I'm building fundamentals strong enough that the frameworks don't matter. I want to understand what's happening underneath — in the data structure, in the database call, in the model's decision boundary.

I want to become someone people can rely on — not because I have all the answers, but because I know how to find them. That standard drives me more than any deadline ever could.

Good code doesn't prove you're smart. It proves you cared enough to think twice.