Company DescriptionZynnexAI builds voice AI infrastructure for modern BFSI institutions, enabling banks, NBFCs, fintechs, insurers, and AMCs to deliver intelligent, in-app voice relationship managers and conversational AI agents. Our platform powers voice-driven customer journeys across KYC, loans, credit cards, insurance, and debt collection. From real-time speech interactions and multilingual conversations to intelligent decision-making and automated customer engagement, we build AI systems that make financial services more accessible, efficient, and conversational. We are building an end-to-end voice AI stack encompassing Speech-to-Text (STT), Text-to-Speech (TTS), LLMs, decision models, agent orchestration, and telephony infrastructure, with a robust focus on real-time performance, scalability, and production reliability. Role DescriptionAs a Founding AI Engineer at ZynnexAI, you will work on building and optimizing the core AI infrastructure powering our conversational and voice AI products. The role is particularly suited for an engineer interested in speech technologies, LLMs, and applied AI research, who enjoys taking models from experimentation to production. You will work across the AI stack, including STT, TTS, LLMs, fine-tuning, model evaluation, benchmarking, inference optimization, and deployment. You will experiment with open-source models, explore different architectures, build training and evaluation pipelines, and optimize models for low-latency, real-time voice interactions. You will also work on intelligent decision-making systems that enable AI agents to understand context, determine appropriate actions, and execute complex workflows. This is a hands-on founding team role where you will have ownership of AI research, model selection, experimentation,



architecture, and production deployment.

Location:
Bengaluru Work Mode: Full-time, On-site

Key Responsibilities
• Voice AI Engineering: Work with STT, TTS, voice activity detection, speech processing, and real-time conversational pipelines.
• LLM Engineering: Experiment with open-source LLMs, prompt engineering, fine-tuning, quantization, and model adaptation for domain-specific use cases.
• Model Development: Build and optimize neural networks, NLP pipelines, classification models, and intelligent decision-making systems.
• Model Benchmarking: Evaluate models across accuracy, latency, TTFT, throughput, resource utilization, and inference cost.
• Training & Fine-tuning: Develop data preparation, synthetic data generation, training, fine-tuning, and evaluation pipelines.
• Deployment & Optimization: Deploy models on GPU infrastructure, optimize inference, and build scalable model-serving systems.
• Agent Intelligence: Develop context-aware decision models and reasoning workflows that enable AI agents to take appropriate actions.
• Research & Experimentation: Continuously evaluate emerging AI models, architectures, frameworks, and techniques to improve our AI stack. Qualifications
• Strong foundation in Computer Science, Machine Learning, Deep Learning, and Data Structures & Algorithms.




• Strong programming proficiency in Python and experience with frameworks such as PyTorch or TensorFlow.
• Practical understanding of LLM architectures, Transformers, embeddings, tokenization, and inference.
• Interest or experience in voice AI technologies, particularly STT, TTS, speech processing, and conversational systems.
• Experience experimenting with open-source LLMs, fine-tuning, quantization, or model optimization.
• Understanding of model evaluation, benchmarking, and performance metrics.
• Familiarity with GPU-based model deployment, inference frameworks, and serving technologies.
• Experience building APIs and integrating ML models into production applications.
• Ability to independently research, experiment, troubleshoot, and take AI systems from proof of concept to production. Good to Have
• Experience with frameworks and tools such as Hugging Face, Transformers, vLLM, Ollama, ONNX Runtime, and TensorRT.
• Exposure to speech frameworks and technologies such as Whisper, NeMo, Indic speech models, LiveKit, and real-time audio pipelines.
• Experience with multilingual LLMs, Indian languages, speech synthesis, and voice cloning.
• Familiarity with model quantization techniques such as INT8, INT4, and FP8.
• Experience with GPU infrastructure, distributed inference, and model serving at scale.
• Experience building conversational AI agents, decision-making systems, or agentic workflows.
• Exposure to BFSI, fintech, or other regulated domains.
• We value strong fundamentals, practical experimentation, and the ability to build things from scratch over familiarity with any particular framework or tool.

📌 Founding AI Engineer – Voice AI (Bengaluru)
🏢 ZynnexAI
📍 Bengaluru