Building Low-Latency Voice Pipelines for Real-Time AI Agents
Exploring the architecture behind sub-200ms voice processing pipelines and the challenges of real-time telephony AI systems...
Read More →> initializing_profile_
$ cat summary.txt
At Voicing AI, my focus is on leveraging Generative AI to transform B2B AI telephony, enhancing customer interactions with AI-enabled voice solutions. We are streamlining communication processes with innovative AI audio telephony, building upon a strong foundation in MLOps and prompt engineering.
Previously at Pixis, the team and I developed a Generative AI platform for Ad Creative generation, utilizing Computer Vision and a RAG based LLM architecture. This experience, along with my proficiency in multiple programming languages, has been instrumental in my current role, driving advancements in AI voicing infrastructure.
$ _
Pixis develops codeless AI infrastructure to help brands scale all aspects of their marketing in a world of complex consumer behavior.
Spearheaded feature development and optimization across microservices in Python, Node, and Go, deployed across Google Cloud, Azure, and AWS.
Backend Developer for the Spoken Tutorial project.
Backend developer for the NVLI Project at IIT Bombay.
Coursera / deeplearning.ai
Coursera / deeplearning.ai
IBM (DL0110EN)
Google / Coursera
Computer Science
2014 — 2017Exploring the architecture behind sub-200ms voice processing pipelines and the challenges of real-time telephony AI systems...
Read More →A practical comparison of Retrieval Augmented Generation and fine-tuning approaches for enterprise LLM deployments...
Read More →Lessons learned from migrating from Google Pub/Sub to Kafka for high-throughput ML inference pipelines...
Read More →