Full-stack engineer · Patna, India
I build fast APIs, sharp interfaces & models that actually learn.
I'm Priyanshu, a computer science engineer from NIT Patna who likes the whole stack. I design the schema, ship the secure API, and make the interface feel effortless. Off the clock, I teach machines to read between the lines and keep my DSA streak honest.
Amazon ML Challenge 2024Ranked around #630 out of about 70,000 participants nationally.
DSA problems solvedOn LeetCode and GeeksforGeeks, and counting.
Model accuracySentiment classifier on a 10,000+ Amazon review test set.
Smaller filesLossless text compression with a hand-built Huffman coder in C++.
Things I've built, broken, and rebuilt better, from AI assistants to compression algorithms.
An AI scheduling assistant you can just talk to.
Type what you want in plain English and TailorTalk turns it into Google Calendar actions: it creates, updates and manages events. Google Gemini handles the language understanding, and each parsed intent is validated before it reaches the Calendar API.
A full-stack membership portal that writes its own certificates.
A membership portal that generates certificates automatically, which cut manual processing time. It uses Firebase Auth for role-based sign-in and Firestore for real-time data, with a mobile-first interface built in React and Bootstrap.
A model that reads Amazon reviews and gets the mood right.
A logistic regression classifier that reached 92% accuracy on a test set of more than 10,000 reviews. It runs on a Pandas and NLTK pipeline (tokenisation, stop-word removal, TF-IDF vectorisation) and was benchmarked against Naive Bayes.
Lossless compression in C++, with the data structures built by hand.
A file compressor built on a custom Huffman coding algorithm. A min-heap and binary-tree structures build the frequency table and the optimal prefix codes. Text files come out 40–60% smaller, and decompression restores every byte exactly.
Internships where I built real products for real users, on the web and in machine learning.
The tools I use every day, grouped by the problems they solve.
as they grow.
RESTful services in C# with ASP.NET Core Web API, data access through Entity Framework Core and LINQ, and Node.js when JavaScript is the better fit.
obvious.
Single-page apps in Angular with RxJS and reactive forms, React apps that stay fast and mobile-first, and HTML and CSS that work on every screen.
honestly.
From raw text to a trained model: cleaning and vectorising with Pandas and NLTK, training and benchmarking with Scikit-learn and TensorFlow, and plotting the results with Matplotlib.
lives.
Relational data in MySQL and SQL Server, real-time data in Firebase Firestore, Microsoft Azure for the cloud, and API contracts checked in Postman and Swagger.
Languages I think in
C#TypeScriptJavaScriptPythonC++SQL