Research & Engineering
Contributions.

Our research, technical whitepapers, engineering insights, and contributions to the field of autonomous intelligence.

Ragini Pro: A High-Fidelity Voice Model for English and Hindi

Udeshya RajAssociate Data Scientist

Ragini: A High-Fidelity Multilingual Voice Model for Indic Speech

Udeshya RajAssociate Data Scientist

Akshara: State-of-the-Art Speech Recognition for Indic Languages

Pavan TaddiAssociate Data Scientist

Ragini Lite: 100× Real-Time Code-Switched TTS for Scale & Edge

Udeshya Raj, Anuj VishwakarmaAssociate Data Scientist

Unveiling Rawi: A Unified Neural Voice Model for Multi-Dialect and Code-Switched Arabic Speech

Amina AshrafData Scientist

Introducing Alif: Next-Generation Speech-to-Text Engineered for the Nuances of Spoken Arabic

Amina AshrafData Scientist

Optimising Autoregressive Speech Models

Udeshya RajAssociate Data Scientist

MSOCA: Multi-Session Omni-Channel Agents

Sneh ShahML Engineer

Engineering the Watchtower

Rohan DivakarML Engineer

The Production Landscape of MCP Gateways

Prasoon PriyadarshiAssociate Data Scientist

Acoustic Fusion — Does adding the agent's audio as context really help?

Pavan TaddiAssociate Data Scientist

Benchmarking SOTA Speaker Diarization Models

Nishant SinhaAssociate Data Scientist

Smart Prompting: what the model actually needs

Niraj ItaliyaML Engineer

Anyone Can Code, But Who Can Architect? — ConvoZen ResearchArchitecture as a graph of decisions

Gokul PradeepData Scientist

Benchmarking Practices, Datasets, and Metric Gaps for Conversational TTS

Boneshwar V KAssociate Data Scientist

Agentic Chunking: The Silent Ceiling on RAG Accuracy

Arjun BhammarML Engineer