Barathi Ganesh Hullathy Balakrishnan

Barathi Ganesh Hullathy Balakrishnan

1st-year doctoral student

Advisor: Prof. Ptaszynski Autumn enrollment

About Me

I build bias-aware, edge-ready speech and language models for under-represented languages. Currently a Ph.D. candidate at the Kitami Institute of Technology , I am developing unified Multimodal, Multilingual, and Multitask (M3) transformers to ensure AI systems serve diverse linguistic identities with demographic fairness and structural equity.

Research Interests

  • Multimodal AI
  • Natural Language Processing, Representation, and Understanding

Background

Period Role / Degree Institution / Company
2025 — Ph.D. in Multimodal AI Kitami Institute of Technology
2020 — 2025 Head of Product Development RBG.AI
2018 — 2020 Chief Technology Officer - R&D Product Manager Arnekt Solutions
2016 — 2017 Research Scientist Analyst Accenture Solutions
2015 — 2016 Assistant System Engineer TATA Consultancy Services
2013 — 2015 M.Tech in Computational Engineering and Networking Amrita Vishwa Vidyapeetham
2009 — 2013 B.E. in Electronics and Communication Engineering CSICE, Anna University

Publications & Achievements

  • AURA-ST: Acoustic-Unconstrained Residual Architecture for Speech Translation
    • Focus: Introduces a novel residual architecture designed for speech translation that operates without rigid acoustic constraints, improving flexibility and translation robustness.
  • KIT-TIP-NLP at MultiPride: Continual Learning with Multilingual Foundation Model
    • Focus: Explores continual learning strategies applied to multilingual foundation models, developed for the MultiPride shared task/benchmark.
  • RBG-AI: Benefits of Multilingual Language Models for Low-Resource Languages
    • Focus: Investigates the impact and performance gains of leveraging large-scale multilingual language models specifically for underrepresented and low-resource languages.
  • RBG-AI at FadeIT: Prompted LLMs with Label Abstraction for Logical Fallacy Detection
    • Focus: Demonstrates the use of prompt engineering and label abstraction techniques with Large Language Models to identify logical fallacies, submitted to the FadeIT challenge.

Google Scholar

GitHub

ORCID

Website

Hobbies

  • Tech & Design: Exploring edge AI deployment, managing digital domains, and designing custom mechanical hardware.
  • Physical Pursuits: Mountaineering and engaging with the outdoors to counterbalance intensive digital research.
  • At-Home Projects: Cultivating plants through hydroponics and unwinding through cooking.