AI SME-Architect ( w / AWS)

AI SME/Architect- HYBRID

Auburn Hills, MI - long term (3 days a week to office)

The Role:

Our Client is seeking a AI SME/Architect with AWS Experience to join our team in Jersey City, NJ (Need Onsite day 1, hybrid 3 days from office).

Our challenge

We are seeking an exceptional hands-on Technical Lead to spearhead our enterprise GenAI engineering program. This is a unique opportunity for a seasoned technologist who combines deep AI/ML expertise with practical engineering skills to build and operationalize cutting-edge generative AI solutions. Candidate will lead the development of AI agents and platforms while remaining deeply involved in the technical implementation.

Responsibilities:

Technical Leadership & Development

• Lead the design, development, and deployment of enterprise-scale GenAI solutions using a hybrid of custom developed solutions and open-source platforms (Dify, OpenWebUI, etc.)
• Architect and implement AI agents using Python frameworks including LlamaIndex and LangGraph
• Drive hands-on development while providing technical guidance to the engineering team
• Establish best practices for GenAI development, deployment, and operations

AI/ML Engineering

• Design and implement LLM-based solutions with deep understanding of model architectures, fine-tuning, and prompt engineering
• Apply classical machine learning techniques where appropriate to complement GenAI solutions
• Optimize AI pipelines for performance, cost, and scalability
• Implement RAG (Retrieval Augmented Generation) patterns and vector databases

Context Engineering & Advanced RAG

• Design and implement sophisticated context engineering strategies for optimal LLM performance
• Build advanced RAG systems including multi-hop reasoning, hybrid search, and re-ranking mechanisms
• Develop agentic RAG architectures where agents dynamically query, synthesize, and validate information
• Implement context window optimization techniques and dynamic context selection strategies
• Create self-improving RAG systems with feedback loops and quality assessment

LLM Optimization & Fine-tuning

• Lead fine-tuning initiatives for domain-specific LLMs using techniques like LoRA, QLoRA, and full fine-tuning
• Implement performance optimization strategies including quantization, pruning, and distillation
• Design and execute benchmark suites to measure and improve model performance
• Optimize inference latency and throughput for production workloads
• Implement prompt optimization and few-shot learning strategies

Platform & Infrastructure

• Design event-driven architectures for asynchronous AI processing at scale
• Build and deploy containerized AI applications using Kubernetes on AWS
• Implement AWS services (SageMaker, Bedrock, Lambda, EKS, SQS, SNS, etc.) for AI workloads
• Establish CI/CD pipelines for AI model and application deployment

Security & Governance

• Implement secure design principles for AI systems including data privacy and model security
• Establish AI security frameworks covering prompt injection prevention, model access controls, and data governance
• Ensure compliance with enterprise security standards and AI ethics guidelines
• Design audit trails and monitoring for AI system behavior

Requirements:

AI/ML Expertise

• Deep understanding of LLMs: Architecture, training, fine-tuning, and deployment strategies
• Context engineering proficiency: Expert-level understanding of context window management, prompt engineering, and context optimization techniques
• Advanced RAG implementation: Hands-on experience building sophisticated RAG systems with hybrid search, metadata filtering, and agentic capabilities
• Fine-tuning expertise: Proven experience fine-tuning LLMs for specific domains using modern techniques (LoRA, PEFT, etc.)
• Performance optimization: Track record of optimizing LLM inference for latency, throughput, and cost
• Classical ML proficiency: Strong foundation in traditional machine learning algorithms and applications
• Python mastery: Expert-level Python with extensive experience in ML libraries (PyTorch, TensorFlow, Pandas, NumPy)
• GenAI frameworks: Hands-on experience with LlamaIndex, LangChain, LangGraph, or similar frameworks
• Open-source GenAI platforms: Experience with Dify, OpenWebUI, or comparable platforms
• Engineering Excellence
• Cloud architecture: Proven experience designing and implementing AWS solutions using multiple services
• Event-driven systems: Expertise in asynchronous, event-driven architectures for scalable AI processing
• Containerization: Advanced knowledge of Docker, Kubernetes, and container orchestration
• DevOps/MLOps: Experience with CI/CD, infrastructure as code, and ML model lifecycle management
• Security & Enterprise Standards
• Secure development: Strong understanding of secure coding practices and security design patterns
• AI security: Knowledge of AI-specific security concerns (adversarial attacks, data poisoning, prompt injection)
• Enterprise integration: Experience with enterprise authentication, authorization, and compliance requirements

Leadership & Communication

• 12+ years of hands-on technical experience with at least 5 years in AI/ML
• Proven track record of leading technical teams while remaining hands-on
• Excellent communication skills to articulate complex technical concepts to diverse stakeholders
• Experience working in enterprise environments with multiple stakeholders

Preferred Qualifications

• Experience with multi-agent systems and agent orchestration
• Knowledge of vector databases (Qdrant, OpenSearch, pgvector)
• Expertise in embedding models and semantic search optimization
• Contributions to open-source AI/ML projects
• Experience with model quantization and edge deployment
• Knowledge of graph-based RAG and knowledge graph integration
• Certifications in AWS, Kubernetes, or ML platforms

Please ensure that you use the below template forma when submitting profiles. Only the following details along with the resume should be shared:

Do not submit any personal documents along with the profile.

Please always reply on the same email thread and keep all point-of-contacts (POCs) in CC.