TeLLaMA
Telugu Language Model — 35M parameter GPT trained from scratch on CC-100
10 200
0 1.5
0.1 1
1 2
Example Prompts / ఉదాహరణలు
TeLLaMA • 35.3M params • 6-layer decoder-only Transformer • BPE-32k • Trained on CC-100 Telugu • GitHub • Model
About the Developer
Anirudh Marpally

Anirudh Marpally

M.Tech (Modelling & Simulation) • Java Developer • ML Researcher

Software engineer with 4+ years of experience building enterprise-grade Java applications at Concentrix Catalyst (Union Pacific Railroad). Specialised in designing secure, scalable microservices with Spring Boot, modernising legacy codebases from Java 8 to Java 17, and maintaining production-grade CI/CD pipelines with 95%+ test coverage. Currently pursuing M.Tech at Defence Institute of Advanced Technology (DIAT), Pune, with research focus on NLP and language modelling for low-resource languages. Passionate about bridging the gap between software engineering best practices and cutting-edge machine learning research.

Java 17 Spring Boot PyTorch Python Transformers NLP REST APIs SQL Git Jenkins
Spring Data JPA Hibernate Microservices JUnit 5 Maven SonarQube JIRA CyberArk MATLAB Simulink Hugging Face C React JS Oracle DB MySQL Agile/Scrum Power Systems Smart Grids Control Systems Numerical Methods Docker
LinkedIn GitHub Hugging Face

Education

M.Tech — Modelling & Simulation, DIAT Pune (CGPA: 7.07)
B.Tech — Electrical Engineering, Vardhaman College (CGPA: 7.87)

Experience

Junior Software Engineer @ Concentrix Catalyst
Since Mar 2022, Java Developer

Research Focus

Low-resource NLP • Telugu language modelling • Reasoning-based LLMs for mathematics • Transformer architectures

Academic & Competitive Credentials

GATE 2022 EE: 357 • GATE 2022 EC: 333
Presented research paper titled “Regenerative Braking Systems in Electric Vehicle” published at RTPEE–2022, SSN College of Engineering

Projects

Reasoning-Based LLM for Mathematical Understanding Ongoing
Oct 2025 – Present

Developing a reasoning-based language model to solve mathematical problems up to 12th-grade level. Focus on symbolic reasoning, logical inference, and step-wise mathematical modelling.

T-LLM: GPT-Style Architecture for Low-Resource Language (Telugu) Ongoing
Aug 2025 – Present

Implementing and fine-tuning transformer-based language models using PyTorch. Focus on tokenization, dataset preprocessing, and architecture scaling for low-resource languages.

Cineworld — Movie Ticket Booking App Completed
Nov 2022 – Feb 2023

Full-stack movie ticket booking application using Spring Boot (backend) and React JS (frontend). Flexible movie selection filters by time, date, and location with Oracle/MySQL integration.

Educational Center Courses Management Completed
Aug 2022 – Sep 2022

Course management system using Spring Boot and Spring Data JPA. CRUD operations for courses, instructors, and students with optimised Oracle queries.

Regenerative Braking System of Electric Vehicle Completed
Jun 2021 – Mar 2022

Modelled and compared regenerative braking efficiencies using synchronous motor, BLDC motor, and induction motor. Developed mathematical models for energy recovery and system performance analysis.

Solar Powered Mobile Charger using Buck Converter Completed
Nov 2020 – Apr 2021

Designed a real-time executable solar-powered charging system using DC–DC buck converter. Targeted portable power solutions for rural and hilly regions with limited grid access.

I'd love to hear from you — whether it's a collaboration opportunity, technical discussion, or feedback on this project. Let's connect!

marpallyanirudh@outlook.com +91 7993124879 LinkedIn