I am an MCA student (graduating 2027) who got tired of tutorials that stop at "it works" and started asking "what happens when it does not." That question is basically my whole engineering personality now - I care less about adding a feature and more about what breaks it, what happens under bad input, and whether the next person touching the code will understand why it is built the way it is.
Most of what I know about backend engineering, I learned by building a full production-shaped application - authentication, validation, error handling, async state, all of it - instead of a CRUD demo. What I wanted to test was not just whether an AI feature works, but whether it works reliably enough to ship.
I am a fresher on paper. I do not think of myself as one in practice - I just have not had a team to prove it to yet.
AI-Powered Web Search & Email Assistant
The question I was chasing - can an AI application search the live web, actually understand what it finds, and turn that into something useful, like a drafted email, without a human stitching the steps together by hand?
User -> React + Redux -> REST API -> LangChain <-> Tavily (live search)
|
|-- JWT + Bcrypt auth, email verified via Nodemailer
|-- Zod-validated endpoints, centralized error handling
+-- MongoDB
Wiring LangChain to a search API was the easy part. The harder part was making the rest of the system boring - predictable auth, validated input, errors that fail loudly instead of silently. That is the part I actually enjoyed building.
Full stack details and feature breakdown - resume - repository
Hackathons, leadership roles, and education are in the resume - this page is the story behind the bullet points.
Full Stack Development - Backend Engineering - AI Integration