SOFT's client located in Philadelphia, PA is looking for an AI Engineer - MCP , LLM , Agentic for a long term contract assignment.
Role Summary
We are looking for a hands-on AI Engineer to design, deploy, and scale intelligent systems leveraging Model Context Protocol (MCP), LLM-powered automation, and agentic AI. This role focuses on building secure, compliant, and testable architectures within strict SaaS constraints and zero public internet egress.
As this technology evolves rapidly, responsibilities may adapt to align with emerging priorities.
Education: Bachelor’s or Master’s degree in Computer Science, AI, Data Science, Machine Learning, or related field OR equivalent professional experience
Experience:
o 3+ years in AI/ML or software engineering
o 1 year setting up and operating MCP servers
o 1 year developing and deploying bots to optimize business processes
Technical Skills:
o Strong Python skills; familiarity with AI/ML frameworks (PyTorch, TensorFlow)
o MCP architecture expertise (providers/tools, resource servers, authentication/secrets)
o Hands-on experience with GenAI and agentic AI systems
o RAG implementation
o AWS operations (API Gateway, Lambda, S3, OpenSearch, CloudWatch, KMS) or equivalent experience in Azure/Google Cloud
o Robotic Process Automation (RPA) development experience
o CI/CD pipelines with containerization (Docker/Kubernetes)
o Structured logging and compliance-driven design
Preferred Qualifications
• Experience mapping prototypes to cloud architectures
• Familiarity with OpenSearch Serverless in GovCloud
• Certifications: AWS ML Specialty, AWS Security Specialty
• Prior work in air-gapped environments with zero public egress
• Ability to design and implement LLM-based bots with robust dialog management and context handling
• Hands-on UiPath development experience
• Strong interpersonal skills for collaboration across internal and external stakeholders
Key Responsibilities
MCP Server Platform (30%)
• Deploy, configure, and manage MCP servers (tools/providers, resource access, authentication/secrets, observability).
• Integrate MCP with internal services and APIs while ensuring compliance with GovCloud security standards.
Conversational AI (10%)
• Build Retrieval-Augmented Generation (RAG) pipelines using GovCloud-compatible services (e.g., OpenSearch Service) for grounded responses.
Agentic AI & Orchestration (20%)
• Develop agentic workflows for task decomposition, planning, tool use, and multi-agent coordination.
• Implement AWS Bedrock guardrails, fallback strategies, and audit trails for safe execution.
Quality Engineering & Testability (10%)
• Extend modern test frameworks (e.g., Playwright) to automate UI and API testing using GenAI capabilities.
• Integrate MCP tools and agentic AI into test automation for dynamic test generation, intelligent validation, and adaptive error handling.
• Ensure structured logging, correlation IDs, and compliance-ready audit records for all test executions.
GovCloud Architecture (20%)
• Map prototypes to AWS GovCloud managed services (API Gateway, Lambda, S3, OpenSearch, CloudWatch, KMS).
• Deliver solutions fully compatible with AWS GovCloud to maintain zero egress.
Security & Compliance (10%)
• Apply least-privilege IAM, KMS-backed secrets, and token scoping for MCP tools.