Conversational AI Engineer - Google/GECX & Code Agents

TTEC Digital; Colombia; Presencial; COP 60.000.000 - 110.000.000; Jornada completa

Descripción de la vacante

TTEC Digital in Colombia is seeking a Conversational AI Engineer specializing in Gemini Enterprise for Customer Experience to design, deploy, and optimize multi-agent workflows that bridge discovery with front-line operations.

You will combine low-code CX Agent Studio with custom Python/TypeScript tooling to build scalable agent networks, integrate with CRMs and CCaaS, and drive automated testing to prevent hallucinations before production.

Formación

• Bachelor’s degree in computer science, engineering field, or equivalent practical experience.
• 3+ years of experience building conversational AI solutions or programmatic AI workflows.
• Proficiency in Python or TypeScript/Node.js with JSON payloads and RESTful APIs.
• Experience with Git-based workflows and CLIs for deployment.
• Hands-on experience with Google GECX Code Agent tooling (cxas-scrapi) and related SDKs.
• Familiarity with Dialogflow CX or CX Agent Studio.
• Experience deploying streaming, low-latency A2A voice agents with telephony/CCaaS integration.
• Knowledge of prompt engineering safety and fallback hooks.
• Experience with BigQuery or Google Cloud Logging to extract transcripts and metrics.

Responsabilidades

• Develop Multi-Agent Architectures: Build modular, hierarchical agent networks using CX Agent Studio and Code Agents.
• Write Custom Code Tools & Callbacks: Implement Python/TypeScript tool callbacks for advanced agent behavior.
• Leverage GECX CLI & Tooling: Run automated workflows and perform linting and semantic checks.
• Design Generative Playbooks & Control Flows: Create structured, XML-like agent instructions with routing syntax.
• Program Enterprise Integrations: Connect Code Agents to databases, CRMs, and CCaaS via APIs/WEBHOOKS.
• Automate Quality & Evaluation (CI/CD): Develop golden-scenario tests and simulations to prevent regressions.

Conocimientos

Conversational AI

Python/TypeScript

RESTful APIs

Git workflows

CLI tooling

Dialogflow CX

CX Agent Studio

A2A voice tech

BigQuery

Cloud Logging

Educación

Bachelor’s degree in computer science, engineering field, or equivalent practical experience

Herramientas

cxas-scrapi

cxas lint

cxas llm-lint

Descripción del empleo

At TTEC Digital, we coach clients to ensure their employees feel valued, and fully supported, because an amazing customer experience is an employee first process. Our vision is the same, a place where employees know they can thrive.

As a Conversational AI Engineer specializing in Gemini Enterprise for Customer Experience (GECX), you will design, deploy, and optimize next-generation, customer-facing agentic workflows. You will work on a unified platform that bridges the gap between digital discovery (via Google Search and Maps) and front-line customer operations (chat, voice, and retail/commerce touchpoints).

You will use Customer Experience Agent Studio (CX Agent Studio) to build multi-agent applications, but your core focus will be implementing Code Agents—leveraging programmatic SDKs (such as cxas-scrapi), writing custom Python/TypeScript tool callbacks, and utilizing localized CLI workflows to programmatically structure, lint, test, and sync complex agent hierarchies. You will merge the speed of low-code visual canvas building with the absolute determinism of custom-coded logic.

What you'll be doing:

• Develop Multi-Agent Architectures: Build and configure modular, hierarchical agent networks (Root/Steering agents and targeted Sub-agents) using a combination of the CX Agent Studio canvas and programmatic Code Agents.
• Write Custom Code Tools & Callbacks: Program specialized backend logic, data transformation scripts, and custom tool callbacks in Python or TypeScript to extend agent capabilities beyond basic natural language prompts.
• Leverage GECX Engineering CLI & Tooling: Run automated code-first developer workflows utilizing uv environments and the cxas CLI for structural static linting (cxas lint) and advanced semantic instruction validation (cxas llm-lint).
• Design Generative Playbooks & Control Flows: Author and maintain structured, XML-like agent instructions (role, constraints, taskflow) that natively hook into custom Python tools ({@TOOL: custom_script}) and sub-agent routing syntax ({@AGENT: specialized_subagent}).
• Program Enterprise Integrations: Seamlessly hook Code Agents into live databases, product catalogs, CRMs, and CCaaS networks (e.g., Genesys, Avaya) via Model Context Protocol (MCP) or secure OpenAPI webhooks.
• Automate Quality & Evaluation (CI/CD): Script programmatic, golden-scenario test cases and simulation evals using simulation frameworks to test multi-turn agent trajectories, preventing hallucinations and conversational regressions before sync-to-prod.

Skills and experience you will bring:

• Bachelor’s degree in computer science, engineering field, or equivalent practical experience.
• 3+ years of experience building conversational AI solutions, virtual agents, or programmatic AI workflows.
• Strong proficiency in Python or TypeScript/Node.js, with hands-on experience handling complex JSON payloads, RESTful APIs, and cloud-hosted webhooks.
• Experience utilizing Git-based developer workflows and command-line interfaces (CLIs) for software versioning and deployment.
• Hands-on experience building with Google’s GECX Code Agent tooling (e.g., cxas-scrapi workspace/SDK, Python-based agent foundries).
• Good technical understanding of Dialogflow CX, CX Agent Studio, or similar advanced, stateful conversational frameworks.
• Experience deploying streaming, low-latency audio-to-audio (A2A) voice agents and integrating with telephony (SIP/RTP) or CCaaS partner environments.
• Solid grasp of prompt engineering safety protocols, blocklists, and programmatic fallback behaviors (e.g., before_model_callback or after_tool_callback hooks).
• Experience with BigQuery or Google Cloud Logging to programmatically extract conversation traces, search text transcripts, and build data-driven performance metrics.