Legal Soft is hiring a Claude AI Agent Developer to design, build, and deploy a suite of 20 autonomous AI agents that replicate the functions of our Marketing department staff. This is a hands-on engineering role requiring deep proficiency with the Anthropic Claude API, multi-agent system design, and tool-use orchestration. You will work directly with the Head of Marketing to understand each staff role, translate those responsibilities into agent logic, and deliver production-ready agents that automate our marketing operations end-to-end.
Key Responsibilities:
AI Agent Architecture & Development
- Design and build 20 role-specific AI agents using the Anthropic Claude API (claude-sonnet / claude-opus) that replicate the day-to-day tasks of Marketing department staff.
- Implement multi-agent orchestration — defining how agents communicate, hand off tasks, and collaborate across workflows.
- Build tool-use integrations enabling agents to take real actions: querying HubSpot, sending reports, analyzing data, drafting content, and triggering workflows.
- Develop agent memory and state management strategies so agents maintain context across sessions and tasks.
Role Replication & Staff Shadowing
- Interview and shadow current Marketing team members to document their responsibilities, decision logic, and outputs.
- Translate documented workflows into agent system prompts, tool definitions, and behavioral constraints.
- Ensure each agent accurately replicates the quality, tone, and judgment of the staff member it is designed to replace or assist.
Tool & Integration Development
- Build and maintain API integrations between Claude agents and core platforms: HubSpot CRM, internal reporting systems, and communication tools.
- Develop custom tools and function-calling schemas that agents use to retrieve data, execute tasks, and escalate decisions.
- Implement web search and retrieval-augmented generation (RAG) where agents require real-time or document-based knowledge.
Testing, Evaluation & Quality Assurance
- Design evaluation frameworks to measure agent accuracy, task completion rates, and output quality against human-staff benchmarks.
- Identify and mitigate hallucinations, off-task behavior, and edge-case failures across all 20 agents.
- Run structured UAT (user acceptance testing) cycles with Marketing stakeholders before deploying each agent.
Prompt Engineering & Optimization
- Write and iterate on sophisticated system prompts that define each agent’s persona, scope, tools, escalation paths, and output format.
- Benchmark Claude model tiers (Haiku, Sonnet, Opus) per agent to optimize the balance of speed, cost, and reasoning quality.
- Maintain a centralized prompt library and versioning system for all 20 agents.
Deployment & Documentation
- Deploy agents in a scalable, monitored production environment with logging, alerting, and performance dashboards.
- Produce clear technical documentation for each agent: architecture diagrams, tool schemas, prompt rationale, and operational runbooks.
- Train Marketing coordinators on how to interact with, monitor, and escalate issues across the agent suite.
AI Integration & Team Adoption
Owning tool adoption across the team is a non-negotiable part of this role. AI tools that go unused do not count as integrated.
- Build, maintain, and enforce SOPs that embed AI tools into daily, weekly, and monthly marketing workflows for every team member
- Conduct regular audits of tool usage across the team — identify gaps, coach laggards, and escalate blockers
- Create how-to documentation, short training guides, and prompt templates so team members can execute AI tasks independently
- Act as the internal AI marketing resource — evaluating new tools, running pilots, and making adoption recommendations
- Track and report on AI adoption metrics: tasks completed via AI vs. manually, time saved, and output volume
- Coordinate with any technical, IT, or operations stakeholders to ensure AI tools are securely and reliably integrated
Requirements
Must-Have
- 3+ years of software development experience with demonstrated AI/LLM integration work.
- Hands-on experience building applications with the Anthropic Claude API or equivalent LLM APIs (OpenAI, Gemini).
- Strong understanding of agentic AI patterns: tool use, function calling, multi-agent orchestration, memory, and RAG.
- Proficiency in Python and/or JavaScript/Node.js for backend agent development.
- Experience integrating LLM agents with real-world APIs and CRM systems (HubSpot experience a strong plus).
- Ability to independently interpret staff workflows and translate them into agent logic without heavy hand-holding.
- Strong written English communication for documentation and async remote collaboration.
Nice-to-Have
- Experience specifically with the Anthropic SDK and Claude model family.
- Familiarity with LangChain, LlamaIndex, or similar agent orchestration frameworks.
- Background in marketing operations, lead generation, or B2B SaaS environments.
- Experience building RAG pipelines or vector-database integrations.
- Portfolio of deployed AI agents or automation tools (GitHub links or demos strongly encouraged).
System and Work Setup Requirements:
- A stable and high-speed internet connection preferably 25 MBPS or higher with backup in cases of power interruption or service provider issues.
- Own PC/laptop with 8GB RAM, and a processor of Core i5 / AMD Ryzen 3 / 2013 Mac equivalent or higher; Windows 10 or above / Catalina 10.15 or newer; with up-to-date web browsers, and security software, 500GB/128 SSD storage
- Headset/earphone with noise cancellation and webcam at 720p or higher resolution
- A designated, quiet, and well-organized workspace free from distractions to facilitate focused work
Benefits
- Permanent remote work setup
- Competitive starting rate paid in USD
- Internet Allowance
- HMO insurance (PH)
- Paid US holidays
- Paid Vacation and Sick Leaves