Senior Data Scientist

We are looking for a high-impact Data Scientist (3-10 years) to build and scale GenAI-powered Voice and Conversational AI systems for real-world enterprise use cases across BFSI. You will work on cutting-edge LLMs, speech (STT/TTS), and multimodal AI, owning the lifecycle from model design experimentation to production deployment. This is a hands-on IC role with strong ownership and direct business impact (sales, collections, customer experience). If you've built production-grade conversational AI systems and enjoy solving latency, accuracy, and real-world deployment challenges, this role is for you.

Responsibilities:

• Build and deploy GenAI-powered voice bots and conversational AI systems.
• Designed end-to-end voice pipelines (STT LLM TTS)with focus on latency and quality.
• Evaluate and optimize LLMs for voice-first use cases (sales, collections, CX).
• Develop RAG pipelines, prompt strategies, and fine-tuned models.
• Improve conversation quality, tool calling accuracy, and response reliability.
• Design conversation workflows, intents, and dialogue systems.
• Work with large-scale conversational datasets (text, transcripts, and audio).
• Mitigate risks: hallucination, bias, safety, and privacy.
• Collaborate with Product, ML Engineering, and Platform teams for production deployment.
• Experiment with multimodal models, agentic workflows, and tool-augmented LLMs.

Key Success Metrics:

• Production deployment of high-quality voice AI systems.
• Improvements in conversation success rate and user experience.
• Low latency, high accuracy, and system reliability.
• Measurable impact on business metrics (sales, collections, CX).

Requirements:

• 3-10 years of experience in NLP/Conversational AI / GenAI.
• Hands-on experience building voice bots or conversational AI systems in production.
• Strong understanding of LLMs (GPT, LLaMA, etc. ), RAG, prompt engineering, fine-tuning, Speech systems (STT/TTS pipelines).
• Experience with deep learning frameworks (PyTorch / TensorFlow / Hugging Face).
• Strong coding in Python + data handling (SQL/NoSQL).
• Experience working with large-scale datasets (text/audio).
• Understanding of latency, scalability, and production constraints.

Good to Have:

• Experience with multimodal AI / audio token-based models.
• Exposure to agentic AI frameworks (LangChain, LlamaIndex, etc. ).
• Cloud ML platforms (AWS SageMaker / Azure ML / GCP).
• Prior experience in BFSI / customer-facing AI systems.