Lucas Jo is a Head of Technology with eight years of experience building production-grade voice AI and speech systems, currently leading LLM-powered VoIP agent and real-time WebRTC dialogue products at Atlas Labs. He combines deep signal-processing roots (M.S. in voice signal processing) with hands-on ML engineering—fine-tuning Whisper/Wav2Vec models to raise call-domain STT accuracy for Korean and Japanese and deploying low-latency inference on Triton/TensorRT. Lucas designs scalable VoIP and STT architectures (AWS EKS, gRPC, Pipecat, LangGraph, Twilio) that handle interruptions, long pauses, and sensitivity control at large traffic volumes. His background in audio hardware and DSP at RADSONE and EXTEGER gives him a practical edge in solving speech quality and latency trade-offs that typical ML practitioners overlook. He has a track record of translating research techniques into reliable products and open-source contributions (e.g., Kaldi Zeroth Korean recipe) that improved real-world ASR performance.
8 years of coding experience
7 years of employment as a software developer
Master of Science (M.S.), Voice signal processing, Master of Science (M.S.), Voice signal processing at 서울대학교 (Seoul National University)
Bachelor of Applied Science - BASc, Electrical and Electronics Engineering, Bachelor of Applied Science - BASc, Electrical and Electronics Engineering at Seoul National University
This is now the official location of the Kaldi project.
Contributions:14 pushes, 4 branches in 2 years 6 months
kaldilocation
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.