AI Implementation Engineer
- Department
- Development
- Location
- LatAm
- Type
- Full-time · Mid
- Posted
The AI Implementation Engineer is a client-facing technical role responsible for deploying Claude and AI-powered systems inside our clients’ organizations — companies from dozens to low thousands of employees.
This is a hybrid between a project manager and a full-stack engineer: you’ll lead workshops, architect AI solutions, build them end-to-end, and train client employees to actually use what we ship. You own the client relationship alongside a dedicated Project Manager, and you report directly to the CEO.
This is not a research role. It is a delivery role for someone who already lives inside the AI tooling ecosystem — has built skills and plugins, deployed Claude-based agent frameworks, shipped MCP servers, and can walk into a client’s conference room and explain why a particular AI architecture is the right one.
The role
Solution architecture & implementation
- Design AI-powered solutions for enterprise clients using Claude API, MCP servers, agent frameworks, and full-stack infrastructure
- Build production-grade implementations end-to-end (Next.js, Supabase, Vercel, Claude Code, Cursor)
- Deploy and configure open-source Claude-based agent frameworks (OpenClaw, Hermes Agent, and similar) inside client environments
- Develop custom Claude Skills, plugins, and MCP servers tailored to client workflows
- Build internal tools, Slack bots, and AI agents that integrate with client systems via APIs
- Make architecture decisions and document them so the rest of the delivery team can extend the work
Client-facing delivery
- Lead solution architecture sessions and product demos with client stakeholders (technical and non-technical)
- Deliver training workshops to client employees — from executives learning Claude.ai to engineers learning to build their own MCP servers
- Translate vague enterprise problems into shippable AI solutions
- Manage technical expectations around what AI can and cannot do, with honesty
- Co-own the client relationship with a dedicated Project Manager — you handle the technical and architectural side, the PM handles timelines, scope, and account health
Deployment & enablement
- Roll out Claude.ai, Claude Code, and related Anthropic products across client teams
- Build adoption playbooks: change management, prompt libraries, internal documentation, governance
- Create reusable templates, skills, and components that compound across engagements
- Provide post-deployment support and iterate based on real usage data
Continuous learning
- Stay current with the Anthropic ecosystem, new model releases, MCP spec changes, and the broader agent tooling landscape
- Complete the Claude Certified Architect certification under the LowCode Agency account
- Contribute internal knowledge: SOPs, reusable skills, deployment patterns
- Help raise the bar for the rest of the delivery team
Requirements
Technical skills
- Strong full-stack engineering background — comfortable shipping production apps in Next.js, working with Supabase, deploying to Vercel
- Hands-on experience with the Claude API: function calling/tool use, prompt engineering, streaming, batch processing
- Built and deployed at least one of: custom Claude Skills, plugins, or MCP servers
- Hands-on experience deploying open-source Claude-based agent frameworks: OpenClaw, Hermes Agent, NanoClaw, or equivalent
- Experience with LangChain or comparable orchestration frameworks
- Built Slack bots, Discord bots, or similar conversational integrations
- Working knowledge of REST APIs, webhooks, and third-party integration patterns (Gmail, Google Drive, calendar APIs, CRMs)
- Comfortable with Claude Code and Cursor as primary development tools
- Database fundamentals (Postgres/Supabase) and basic auth/security practices
Anthropic ecosystem fluency (required)
- Completed Anthropic’s Skilljar courses (proof required during interview)
- Deep familiarity with the Model Context Protocol (MCP) — has read the spec and built against it
- Working knowledge of Anthropic’s product surface: Claude.ai (Pro/Team/Enterprise), Claude Code, Claude API, Console
Client-facing skills
- English proficiency: minimum 85 score on the EnglishFirst quick test (non-negotiable — this role lives on calls with U.S. enterprise clients)
- Can lead a workshop without notes
- Can present a technical architecture to a non-technical executive without losing them
- Can disagree with a client respectfully when their idea is the wrong one
- Writes clearly: documentation, follow-ups, technical proposals
- Comfortable on camera (Zoom/Meet) for several hours per day
Experience
- 2–4 years of full-stack engineering experience, with at least the last 12 months focused on AI/LLM-powered systems
- Has shipped at least one production AI feature or product to real users
- Bonus: prior consulting, agency, or solutions engineering experience
- Bonus: open-source contributions to AI tooling
Challenge question
Every candidate completes a take-home challenge. Send your response with your application.
AI Implementation Engineer Challenge: Enterprise Knowledge Management Solution
Scenario. TechFlow Manufacturing (850 employees) has hired LowCode Agency to solve their knowledge management crisis. Their engineering team of 120 people across 3 time zones (US East, US West, Singapore) is drowning in tribal knowledge scattered across:
- 2,400+ Confluence pages (mostly outdated)
- 15,000+ Slack messages per week in
#engineering-help - 200+ Google Docs with “the real process”
- Email chains with critical vendor information
- Senior engineers spending 40% of their time answering repeated questions
Key constraints:
- Budget: $25K implementation + $8K/month ongoing costs maximum
- Timeline: 6-week delivery window (they have a major audit in Q2)
- Security: SOC2 compliant — no data can leave their AWS environment
- Integration: Must work with existing Slack, Confluence, and their custom ERP system (REST API available)
- Adoption: Previous knowledge management initiatives failed due to low adoption
Success metrics:
- Reduce engineering support requests by 60% within 90 days
- 80% team adoption within 30 days of launch
- Sub-3-second response time for knowledge queries
Your challenge
Design and justify an end-to-end AI solution architecture that solves TechFlow’s knowledge management problem. Your solution will be compared against Claude’s baseline recommendation in a blind review.
Deliverables:
1. Solution Architecture (1 page)
- High-level technical architecture with specific tools/frameworks
- Data ingestion and processing pipeline
- User interaction patterns and interfaces
- Deployment approach within their constraints
2. Implementation Strategy (1 page)
- Phased rollout plan with specific week-by-week milestones
- Change management approach addressing their adoption challenge
- Cost breakdown showing how you stay within budget
- At least 2 significant trade-off decisions you made and why
- One contrarian choice that Claude would likely not recommend (and your justification)
Evaluation criteria
Your response should demonstrate:
- Real-world experience with similar enterprise deployments
- Creative problem-solving that goes beyond obvious AI chatbot solutions
- Understanding of enterprise adoption challenges and human psychology
- Technical depth in Claude API, MCP, and agent frameworks
- Ability to balance ideal solutions against real constraints
Time limit: 2 hours maximum. Quality over length.
Show us you can think beyond what AI generates. Use any tools you want — but add your own insight.
How to apply
Click Apply for this role below, or email careers@phosailabs.com with:
- Your resume or LinkedIn
- Your response to the challenge question above
- Proof of your Anthropic Skilljar course completion
We respond to every applicant within 5 business days.