What does a reward model really learn?
Testing the gap between reward-score accuracy, policy agreement, and action correctness in language-conditioned navigation.
Explore project ↗RESEARCHER · EDUCATOR · TECHNICAL LEAD
Understanding models.
Building with
evidence.
I’m Anandaswarup Vadapalli, a machine learning researcher working across language, multimodal AI, and speech. I build models, examine how they behave, and connect evaluation to human experience.
Assistant Professor · MSIT, CIHL, IIIT Hyderabad CampusTechnical Lead (Consultant) · ResearchSpark Hub
RESEARCH FOCUS
LLM & multimodal evaluationEfficient machine learningSpeech & languageA CLOSER LOOK
Selected research
The hidden costs of translation accuracy
Understanding the trade-offs between translation quality, inference latency, and estimated emissions.
Explore project ↗Giving synthetic speech room to breathe
A research thread from word-terminal syllables and learned representations to phrase-aware neural text-to-speech.
Explore project ↗HOW I WORK
A model’s score is
the start of the question.
My work combines controlled experiments, hands-on model development, and human evaluation. I’m interested in what a benchmark captures, what it misses, and how those differences affect the people using a system.
More about my work ↗RECENT WORK
From the publication desk
Accepted · ACDSA 2027
Preprint · 2025
2025
LET’S CONNECT
Good research starts
with a conversation.
Research collaborations, applied ML opportunities, and conversations about evaluation, language, or speech.