AI Agent Architect - Conversational AI Platforms ?

At Shakers, we're partnering with a leader in the conversational AI space, working with large enterprises, to bring on an AI Agent Architect for a new project. Remote role, hands-on with agentic architecture and AI coding tools like Claude Code.

The challenge The company already runs a conversational AI platform live across voice, chat and messaging channels, and is opening new projects this year.

Someone needs to own how those problems get framed: how a router agent hands off to specialists, where the guardrails sit (user recognition, queue management, handoff to a human), and where the line falls between retrieving an answer and saying "I don't know" so agents never invent facts. You'll work closely with Product, DevOps, Customer Success and Technical Support to take flows from design to production, direct the build with tools like Claude Code, and stay hands on after launch to monitor quality and iterate.

Responsibilities

- Frame and design agentic architectures (router plus specialist agents) for chat and voice channels.
- Define guardrails and gates: user recognition, queue management, escalation to a human agent.
- Set the boundaries of each agent: when to answer, when to retrieve information, when to say "I don't know".
- Translate business goals into concrete rules, flows and guardrails.
- Shape tone, messaging and channel adaptation across voice, chat and messaging.
- Direct agentic coding tools (e.g. Claude Code) to build, review and validate agent behavior.
- Monitor performance and KPIs post-launch, manage edge cases and iterate.
- Guide the client through decisions and translate them into implementation.
- Coordinate with Product, DevOps, Customer Success and Technical Support.




- Train and enable the client's team to manage and optimize their AI Agents independently.

Requirements

- Proven experience designing agentic systems end-to-end: framing the problem, deciding on tools and context retrieval, designing flows, and defending the reasoning behind those decisions.
- Track record identifying and preventing hallucinations, with clear criteria for when an agent responds, retrieves, or admits it doesn't know.
- Product and UX oriented mindset: designs from the end user's perspective and bridges business goals with technical solutions.
- Daily, hands-on use of agentic coding tools (e.g., Claude Code) to direct build work and judge whether the output is correct.
- Demonstrated portfolio deploying virtual assistants, conversational agents or AI products.
- Working understanding of APIs and integrations.
- Basics of authentication, privacy and security.
- Reading-level Python.
- Familiarity with GDPR and the EU AI Act.
- Native Spanish and very good English, excellent written and verbal communication.
- Availability for occasional in-person meetings in Madrid.

Nice to have

- Experience with CCaaS platforms or voice assistants.
- No-code/low-code automation (n8n, Make, Zapier).
- Enterprise client experience with tools like Jira, Notion, Zendesk or Figma.
- Italian.

Why this project

- Long-term: 12 months from the start, with room to grow.
- Full-time, 100% remote (occasional on-site meetings in Madrid).
- Real strategic weight: you shape how a growing AI company frames and ships its agent architecture for a brand new market.
- Direct visibility with the client's decision-makers and cross-functional teams (Product, DevOps, CS, Support).

If this sounds like you, or like someone in your network, apply or share it. We'll be in touch soon ?

#ConversationalAI #AgenticAI #ClaudeCode #AIAgents #Freelance #RemoteWork

📌 AI Implementation Strategist (Madrid)
🏢 Shakers
📍 Madrid