Headline

Speech AI @ Jio Platforms Limited | Machine Learning Engineer – Speech AI, NLP | ASR, Speech Translation | Agentic AI

About

Machine Learning Engineer specializing in Speech AI and Multilingual NLP, with experience building Automatic Speech Recognition (ASR), Speech Translation, and Neural Machine Translation systems. At Jio Platforms Limited, I work on developing scalable speech technologies for Indic languages, focusing on multilingual modeling, code-mixed speech recognition, and large-scale training pipelines. My work has contributed to improving Transformer-based ASR systems across 10+ languages and building robust preprocessing pipelines for multilingual speech datasets. My work involves designing and training deep learning architectures for speech and language, including Transformers, Conformer, Whisper, Zipformer, and Branchformer, using frameworks such as PyTorch, K2, Icefall, ESPnet, and OpenNMT. I am particularly interested in: • Multilingual and low-resource speech systems • Speech-to-text and speech translation • Agentic AI systems and autonomous AI workflows Always interested in discussions around Speech AI, multilingual ML systems, and emerging AI system architectures.

Location

Bengaluru

AI stack

claude code

claude

codex

hugging face

langchain

openai

Work experience

Machine Learning Engineer at O-Health, Aug 2026 - Present · 1 mo, Bengaluru, At O-health, I am working on developing end to end ASR pipeline for all the Indic Languages to be served in our products

Associate Data Scientist at Jio Platforms Limited (JPL), Jun 2025 - Jul 2026 · 1 y 1 mo, Bengaluru, Karnataka, India, • Trained and deployed Transformer-based Speech-to-Text (STT) models for 10+ Indic languages, improving multilingual acoustic model accuracy. • Developed Code-Switching and Code-Mixed English + Indic ASR systems for real-world conversational scenarios. • Enabled deployment of multilingual ASR systems in IVR and voice bot applications to improve customer interaction experiences. • Built scalable preprocessing and data engineering pipelines for large-scale speech-text datasets. • Developed text normalization, cleaning, deduplication, and tokenization pipelines for multilingual ASR training data. • Implemented Indic language-aware preprocessing strategies to handle script diversity, noisy annotations, and English–Indic code-mixed data.

Research And Development Intern at Reverie Language Technologies, Aug 2024 - May 2025 · 10 mos, Remote, • Trained Proof-of-Concept Speech-to-Text Translation (S2TT) models for multilingual speech recognition and translation tasks. • Conducted experimentation with deep learning-based speech architectures and evaluated model performance across multilingual datasets.

Cloud and Digital Trainee at PwC Acceleration Centers in India, Feb 2024 - Jun 2024 · 5 mos, Remote, The Launchpad Program exposed me to the topics, making me Industry ready, like • IT Fundamentals - Agile Methodology, Software Development architecture and workflows. • RDBMS Using Oracle • Programming Fundamentals - Java • Devops & Cloud Fundamentals - Version Control System, Azure, GCP, AWS essentials • Responsive Web Designing - HTML, CSS, Javascript, JQuery

Research Intern at Samsung R&D Institute India, Apr 2023 - Feb 2024 · 11 mos, Remote, • Reviewed and analyzed several existing DeepLearning models for audio signal processing. • Utilized Pytorch framework for implementing a state-of-the-art model to generate binaural audio from mono audio enhancing user experience.

Education

Bachelor of Technology - BTech, Computer Software Engineering, Siksha 'O'​ Anusandhan University

St. Paul's School, Rourkela

Public links

github: https://github.com/anandtharad