Principal AI Systems Architect | Real-Time Agent Assist / Speech AI Scale

Title: Principal AI Systems Architect | Real-Time Agent Assist / Speech AI Scale Review

Duration: 3 - 6 Month On-going Contract
Start Date: ASAP
Location: Remote - must sit in the US

Position Overview

We have a highly specialized advisory engagement with Harpin that is not a traditional Contact Center Architect or AI Architect search. The client is looking for someone who has personally built, scaled, or led engineering for a real-time agent guidance, agent assist, or live transcription platform in production and can provide an architecture and scale readiness assessment ahead of a major deployment.

Opportunity Overview

Client is preparing to launch its real-time agent guidance platform across approximately 800 agents, 7 deployment waves, and 6 BPO partners between October and December. The product works today at pilot scale. The objective of this engagement is to validate whether the platform can successfully scale in production before UAT begins on October 10.

This is a fixed-scope, remote consulting engagement (1-3 months) focused on architecture review, scalability, latency, streaming ASR performance, endpoint deployment, and production readiness.

What the Client Really Needs

The product:

• Listens to live customer calls
• Performs real-time transcription
• Analyzes conversations as they occur
• Surfaces guidance and knowledge articles to agents during the conversation

The challenge is not integration. The challenge is everything that happens between: Customer speaks → Audio captured → ASR → Inference → Guidance displayed

The client wants someone who has already solved those problems at scale.

Required Experience

Candidates must have experience with:

• Production-scale Agent Assist, Agent Guidance, Speech AI, or Live Transcription platforms
• Streaming ASR and low-latency architectures
• Real-time processing of concurrent audio streams
• Deploying software into managed enterprise or outsourced desktop environments
• Providing executive-ready findings and recommendations