Veeresh Hiremath

I’m a Speech AI / Applied AI Engineer with 1.5+ years of experience building production-grade Voice AI systems, conversational AI agents, real-time ASR pipelines, telephony AI platforms, and LLM-powered applications.

I’ve shipped end-to-end ASR → LLM → TTS pipelines and optimized streaming inference using NVIDIA NeMo, Whisper, wav2vec2, Parakeet RNNT, Conformer architectures, and GPU acceleration. I focus on low-latency design—VAD, audio chunking, speaker segmentation, and robust benchmarking using WER/latency metrics.

I build scalable AI backends with FastAPI, WebSockets, and microservices, and I integrate telephony infrastructure using Asterisk, LiveKit, and WebRTC for real-time conversational experiences. My work also includes speech data engineering automation for dataset preparation, transcript validation, metadata auditing, audio preprocessing, and QA—so models perform reliably in production.