Forward Deployed Claude Engineer / AI architect

Forward Deployed AI Engineer / AI Architect

The customer, a large enterprise with established AI tool adoption, is undertaking a greenfield transformation to establish a repeatable, AI-first organizational workflow model on the Claude platform. Currently, the client lacks a formal Claude environment, including skill estates, packaging, distribution, and deployment pipelines.

This role requires a single, hybrid practitioner—acting simultaneously as an agent builder and platform engineer—who arrives at Week 0 to lead Phase 1 of the workflow redesign alongside an AI Process Designer. Working directly within client teams, the consultant will build a production-grade agentic workflow while establishing the underlying infrastructure, governance, and CI/CD pipelines needed for the client to independently replicate the model across future workflows.

Responsibilities

Agent & Workflow Engineering (Half A)

• Agent Development: Design and ship production-ready agents on the Claude platform using the Agent SDK, Model Context Protocol (MCP), tool use, and context engineering on managed cloud model infrastructure.
• Data Layer Structuring: Conduct readiness assessments, build classification and routing mechanisms, and develop a semantic layer to make enterprise data assets consumable by LLMs.
• Guardrails & Governance: Engineer, configure, and test hard controls and safety guardrails as code artifacts rather than high-level policy descriptions.
• Enterprise Integration: Integrate agents into core business systems (e.g., planning and pricing, marketing/content stacks, operational maintenance tools) while setting up evaluations, observability, and per-workflow token/spend tracking.
• Sprint Execution: Iterate rapidly against target business workflows in agile sprints directly alongside embedded business stakeholders.

Platform & Environment Architecture (Half B)

• Estate Architecture: Define and defend the structural breakdown between skills, agents, connectors, hooks, and prompts. Separate identity from context, optimize frontmatter descriptions for routing, implement progressive disclosure, and set up rule-based retrieval over similarity-based search.
• Packaging & Versioning: Organize repo structures (skills/, agents/, commands/, hooks/), set up sibling path resolutions, enforce semantic versioning, and treat prompt configuration as code via Git PR workflows and CI/CD validation.
• CI/CD & Evals: Build CI pipelines that validate frontmatter, test asset paths, enforce description limits, run script tests, evaluate skill triggering accuracy, and audit outputs against specifications.
• Distribution & Operations: Wire the organizational marketplace to the source repository for seamless auto-updates across surfaces (Claude Code, Cowork, Chat). Implement telemetry, cost tracking, enterprise policy controls, staged rollouts, and rapid rollback mechanisms.

Qualifications and Technical Skills

• Core AI & Platform Stack: Deep technical mastery of the Claude ecosystem (Agent SDK, MCP, tool integration, context engineering) and cloud-managed LLM infrastructure (AWS Bedrock, Azure OpenAI/Foundry, or GCP Vertex AI).
• Software & Platform Engineering: Strong background in Developer Experience (DevEx), Internal Developer Platforms (IDP), or build tooling. Proficiency with Git workflows, semantic versioning, CI/CD pipeline automation, and codebase modularization.
• Data & Integration Engineering: Hands-on experience structuring data layers for LLMs, semantic modeling, API/connector design, SSO integration, masked data views, and data security/auditability in regulated environments.
• Evaluations & Observability: Experience setting up LLM observability, spend monitoring, usage telemetry, and automated eval loops for agent performance.

Background & Experience:

• Proven track record as a Staff-level Engineer, Platform/DevEx Engineer, or Forward Deployed Consultant at an agent/platform vendor.
• Rare signal: Demonstrated experience running a shared agent environment for an organization.
• Soft Skills & Mindset: Highly consultative, business-facing, and capable of driving technical decisions while co-designing workflows in the room with business users.

Summary of the Practicalities

• Project start time: October 2026
• Project duration: Until the end of the year (High possibility for extension)
• Allocation: 100%
• Location requirements: Hybrid (Customer office in Helsinki; candidate must be currently based in Finland)
• Language requirements: Fluent English

More information -> slack: Juho_Witted // juho.sopanen@witted.com