UST
Website:
ust.com
Job details:
Role Description
Conversational AI & IVR Developer (GCP & GenAI)
Role Overview
This is a technical developer role focused on building and optimizing enterprise-grade
IVR and chat solutions. You will be responsible for developing conversational agents that bridge traditional telephony systems with modern Generative AI using the Google Cloud Platform (GCP) ecosystem.
Core Responsibilities
- IVR & Voice Engineering (Dialogflow CX)
- Conversational IVR Design: Build end-to-end voice flows using Dialogflow CX, focusing on state-based design, event handlers, and multi-turn repair strategies.
- Telephony Integration: Configure and manage telephony gateways (e.g., AudioCodes, Avaya, Genesys, or Twilio) and handle telephony-specific signals like DTMF (keypad input).
- Voice Optimization: Fine-tune Speech-to-Text (STT) and Text-to-Speech (TTS) settings, including speech adaptation, barge-in management, and synthetic voice profiles.
- Call Handling: Implement advanced fulfillment logic for call transfers, placing callers on hold, and retrieving metadata like caller_id from the telephony payload.
- Conversational AI & ADK Implementation
- Agent Orchestration: Use the Agent Development Kit (ADK) to build and deploy goal-oriented agents capable of handling complex, non-linear user journeys.
- Reasoning & Task Logic: Configure agents to use LLMs (like Gemini) as a reasoning engine to determine the best next action, tool call, or response based on real-time user intent.
- Dialogflow CX Integration: Build hybrid architectures that combine state-based flows for deterministic tasks (like payments) with generative playbooks for natural conversation.
- GCP Backend & Integration
- Webhook Development: Write backend services (Node.js or Python) for real-time data fetching, business logic, and session management.
- Data Architecture: Manage session states in Firestore and export conversation logs to BigQuery for intent analysis and IVR performance tracking.
- Connectivity: Integrate agents with CRMs and third-party APIs via Cloud Functions to provide personalized, data-driven caller experiences.
- Responsible AI & Reliability
- Guardrail Frameworks: Implement safety layers to prevent LLM hallucinations, enforce ethical responses, and protect against adversarial "jailbreaking" attempts.
- Automated Voice Testing: Use Python-based frameworks to test hundreds of utterances, specifically identifying voice-path issues like background noise interference or "no-input" timeouts.
Technical Skills Required
- Conversational Platforms: Dialogflow CX (Pages, Flows, Route Groups), Playbook CX.
- IVR/Telephony: SIP/RTP knowledge, DTMF handling, STT/TTS tuning, and telephony gateway configuration.
- Programming: Python or Node.js (specifically for Cloud Functions).
- GCP Stack: Firestore, BigQuery, Cloud Logging (Logs Explorer), Vertex AI.
- AI Models: Gemini 1.5 Pro, Vertex AI Search and Conversation.
Collaboration
- Collaborate with Conversation Designers to translate "Happy Paths" into robust technical state machines.
- Troubleshoot production issues using GCP Logs Explorer to identify bottlenecks in webhook latency or speech recognition errors.
- Support the team by documenting prompt strategies and contributing to the evolution of the IVR architecture.
Skills
python,dialogflow cx,agent development kit,shell scripting,
Click on Apply to know more.