Senior AI Researcher

Revrag Ai; Bengaluru; On-site; INR 2,500,000 - 4,200,000; Full time

Job summary

RevRag AI in Bengaluru is seeking a Senior AI Researcher to advance voice AI models and bring research into reliable production systems. You will lead hands-on work in speech recognition, synthesis, and language understanding, while collaborating with the AI Head and engineering teams.

The role emphasizes real-world impact on BFSI conversations, fast iteration, and setting research standards across the AI team. Experience turning research into scalable products is essential.

Qualifications

• 5-7 years of AI/ML research or applied research.
• Strong experience in Voice AI, including speech recog, TTS, or real-time voice systems.
• Hands-on model fine-tuning and domain adaptation.
• Track record of turning research into production systems.

Responsibilities

• Advance Voice AI models for recognition, synthesis, and understanding.
• Fine-tune and adapt models with techniques like LoRA/QLoRA.
• Design rigorous evaluation metrics (WER, latency, user preference).
• Collaborate with AI and platform engineers to deploy at scale.
• Mentor researchers and set research standards.

Skills

Voice AI

Model fine-tuning

Experiment design

Production deployment

Independent problem solving

Communication with leadership

Education

Bachelor/Master in CS/AI/ML

Tools

Python

PyTorch

GPUs training

Job description

About the role

We are looking for a Senior AI Researcher to advance the models behind RevRag AIs Voice AI agents.

This is a hands-on research role for someone who has built and improved Voice AI systems such as speech recognition, text-to-speech, or real-time voice agents; fine-tuned and adapted models and measured the gains rigorously; and taken research ideas into production where they serve real users.

You will work closely with the AI Head and the AI engineering team to improve the accuracy, naturalness, and latency of our voice agents, fine-tune and evaluate models for multilingual, real-world customer conversations, and turn promising research into reliable production capabilities.

This is a research role with a production focus. We value results that work for real users in real BFSI conversations, not only strong benchmark numbers.

Voice AI Research

• Research and develop approaches to improve speech recognition, speech synthesis, and spoken language understanding.
• Work on conversational behaviour such as turn-taking, interruptions, and low-latency responses.
• Improve performance on Indian languages, accents, and code-switching in noisy, real-world audio.

Model Fine-Tuning Adaptation

• Fine-tune and adapt open-source and proprietary models using techniques such as supervised fine-tuning, LoRA, and QLoRA.
• Build and curate datasets, and design data collection and labelling strategies.
• Make informed trade-offs between model quality, latency, and cost.

Evaluation Experimentation

• Design rigorous evaluation methods and metrics, such as WER, latency, and human preference tests.
• Run well-controlled experiments and document findings clearly.
• Track production quality and identify where models need to improve.

From Research to Production

• Work with AI and platform engineers to deploy and serve models reliably at scale.
• Optimize inference for latency, throughput, and cost.
• Support monitoring, error analysis, and continuous improvement of deployed models.

Research Leadership

• Track new research and models, and judge where they can create product advantage.
• Mentor researchers and interns through reviews and technical discussions.
• Share learnings and help set research standards across the AI team.

Experience

• 5-7 years of experience in AI/ML research or applied research.
• Strong experience in Voice AI, including speech recognition, text-to-speech, or real-time voice systems.
• Strong hands-on experience in model fine-tuning and adapting models for specific domains.
• Track record of turning research into production systems.

Technical Skills

• Strong foundation in machine learning, deep learning, and statistics.
• Strong proficiency in Python and deep learning frameworks such as PyTorch.
• Experience training, fine-tuning, and evaluating large models on GPUs.
• Understanding of audio and speech processing concepts.
• Experience designing experiments and evaluation frameworks.
• Understanding of model inference, latency, and cost considerations.

Leadership Communication

• Ability to work independently on ambiguous, open-ended problems.
• Clear communicator who can explain technical findings to engineers and leadership.
• High ownership and focus on results that work in production.

Education

• Degree in Computer Science, AI/ML, or a related field.

Preferred Qualifications

Experience with any of the following will be highly valued:

• Multilingual speech, Indian languages, and code-switching.
• Real-time voice agents, streaming ASR, or low-latency TTS.
• LLM fine-tuning, preference optimization, or evaluation.
• Model serving and inference optimization.
• Published research or open-source contributions in speech or language modelling.
• Building AI systems for BFSI, fintech, or other regulated industries.

Success in This Role

Youll be successful if you:

• Improve voice agent quality with measurable gains in accuracy, naturalness, and latency.
• Deliver fine-tuned models that perform well on real customer conversations.
• Establish clear evaluation methods and metrics for voice and language models.
• Move research quickly into reliable production capabilities.
• Raise the research bar and help build a strong AI team.

About RevRag AI

RevRag AI is a Bengaluru-based deep-tech startup building production-grade Voice AI and LLM-powered automation for customer-facing workflows. Our multilingual, in-app AI agents plug directly into BFSI journeys, improving digital onboarding, sales, and support while cutting operational cost by up to 70%.

We work with leading BFSI organisations including Mahindra Finance, Motilal Oswal, PayTM Insurance, PhonePe, KreditBee, Razorpay, Udaan Capital, Kissht, PolicyBazaar, InsuranceDekho, Open Capital, and EaseBuzz.

Disclaimer: This job posting has been aggregated from external source. Role details, content, and availability are subject to change. Applicants are advised to confirm the latest information directly on the company website before applying.