$ compiling profile...
Shubham Kumar

Shubham Kumar

M.Tech Computer Science (Spec - Computer Systems) @ IIT Gandhinagar

// about

I am Shubham Kumar, a M.Tech student in Computer Science and Engg. (Computer Systems) at IIT Gandhinagar, working under the supervision of Prof. Abhishek Bichhawat.

Prior to this, I completed my B.Tech in Computer Science and Engg. at Mahatma Gandhi Central University, Bihar.

I am currently a member of the FUSS Research Group.

Alongside my academic work, I serve as a Teaching Assistant for undergraduate courses, for which I have been awarded an institute fellowship.

Area of Interest // my_interest

{ }

Compilers and CPU Parallelism.

Research & Projects // present + past

{ }
2025

Natural Language Processing

Bhasha Detect is a lightweight Transformer model that identifies 22 Indic languages across both native scripts and Romanized text within a single consolidated framework. It uses a custom SentencePiece tokenizer and a two-phase training pipeline—combining Triplet Loss pretraining for script-invariant embeddings with Cross-Entropy fine-tuning for classification. The system achieves 96.95% accuracy on native scripts and an overall benchmark accuracy of 88.71% at a fast throughput of ~330 sentences/sec. It effectively supports low-resource languages (e.g., Santali, Manipuri, Dogri) and is fully deployed as a web application.

Indic Language Identification System.

2026

Distributed Systems

PrimeFinder is a distributed prime-finding system built in Go and gRPC, combining a coordinator-worker architecture with an AFS-inspired distributed file system. Distributed File System: Features whole-file client-side caching, primary-backup replication (quorum of 2), and at-most-once RPC semantics with atomic disk renames for crash safety. Compute Engine: Uses a coordinator to assign 500-line data chunks to workers, which run an overflow-safe 6k (+-) 1 trial division primality test. Fault Tolerance: Uses periodic Chandy-Lamport snapshots and worker heartbeats to enable seamless state restoration across worker crashes or cluster failures.

PrimeFinder: Distributed File System and Prime Finding Application.

2026

Compilers

Working on it, Do come and check later :)

Experience // roles

{ }
Aug 2025 - Dec 2025

Teaching Assistant

Took Lab sessions , track of quizzes , assignments , copy check and many more .

ES 242 : DATA STRUCTURES AND ALGORITHM

Jan 2026 - May 2026

Teaching Assistant

Took Attendances , Lab sessions , track of quizzes , assignments , copy check and many more .

ES 119 : Principles of AI

April 2026 - July 2026

Subject Matter Expert — GeeksforGeeks (GATE CS)

Framing questions for test series , daily practice problems , pyqs solving and many more things that helps in understanding the concepts through questions for the students, glooming the hands-on practices with new types of questions in the course of Computer Architecture and Compiler Design.

Subjects - COA & CD

July 2026 - Dec 2026

Teaching Assistant

Taking Lab Classes .

CS 202 : Software Tools and Techniques for CSE .

Achievements // milestones

{ }
↓ Download CV
Mar 2021

JEE Main

Qualified

Qualified for JEE Advance

Apr 2023

NPTEL

Top 5% Topper

Subject - Literature & Life

Feb 2024

GATE

AIR - 10545/1,23,967 candidate appeared for the exam . Score - 418

Subject - CS & IT

Feb 2025

GATE

AIR - 805/1,70,825 candidate appeared for the exam . Score - 737

Subject - CS & IT

Jun 2025

UGC NET

AIR - 47 and Percentile - 99.88 Junior Research Fellowship(JRF) awarded by MoE.

Subject - Computer Science and Applications

Dec 2025

Bihar STET

Score - 108.0802/150

Subject - Computer science

Feb 2026

GATE

AIR - 1308/2,11,020 candidate appeared for the exam . Score - 689

Subject - CS & IT

Contact // reach me

{ }
LinkedIn ↗ Email ↗

// You can meet me at CSE desk 71 at CSE Workspace , AB 13.