AI Platform Architect

Job Title: AI Platform Architect

We're looking for an AI Architect to lead the design and delivery of end-to-end AI solutions—ranging from problem framing and model selection (LLMs and classical ML) to environment architecture, security, and cost optimization. In this role, you'll work closely with a Solution Architect (focused on broader system and enterprise design), while owning the AI-specific architecture: which models to use, what vector/search technologies to adopt, which AWS services to leverage, and how to integrate with data platforms—all while ensuring scalability, safety, and efficiency.

What You'll Do

• AI Architecture Ownership: Lead architecture for initiatives involving RAG, agents, predictive models, NLP, and computer vision. Define reference architectures, target states, and design patterns (batch/real-time, online/offline inference).
• Model Selection & Evaluation: Choose LLMs/foundation models based on use case—whether through Amazon Bedrock (Anthropic, Mistral, Meta, Cohere, etc.), SageMaker-hosted, or open-source. Define evaluation strategies (quality, latency, safety), set up guardrails, and fallback mechanisms.
• AWS AI/ML Stack Design: Align use cases with AWS services—Amazon Bedrock, SageMaker (Studio, Training, Inference, Model Registry), S3, Lake Formation, Kendra, OpenSearch, Lambda, Step Functions, EKS/ECS, API Gateway, CloudWatch, ECR, Secrets Manager, KMS, MSK/Kinesis, Glue, Athena, Redshift, and PrivateLink/VPC endpoints.
• Vector & Search Tech Evaluation: Assess and standardize vector search solutions (Kendra, OpenSearch vector, MongoDB Atlas Vector, pgvector, Pinecone, Weaviate). Define ingestion flows, embedding strategies, filters, schemas, TTL policies, and operational practices.
• RAG & Agent Design Patterns: Architect retrieval pipelines (chunking, hybrid search, re-ranking), prompt orchestration, tool/function calling, persona/policy layers, safety filters, and caching.
• Environment Planning: Define dev/test/prod environments, including network isolation, data zoning, GPU/accelerator strategy, and CI/CD for both models (MLOps) and prompts (PromptOps). Implement rollout strategies like blue/green and canary deployments.
• Cost & Performance Optimization: Deliver FinOps forecasts and enforce guardrails—budgeting tokens/throughput, autoscaling strategies, quantization, distillation, response caching, and batch vs. real-time tradeoffs.
• Security, Privacy & Governance: Implement safeguards for PHI/PII, encryption (at rest/in transit), access controls, key management, data retention policies, and threat mitigation (e.g., prompt injection). Contribute to model risk documentation.
• ML Platform Integration: Collaborate with ML engineers and data scientists on feature stores, experiment tracking, offline/online consistency, A/B testing, and evaluation pipelines.
• Operational Readiness: Define SLOs/SLIs, build observability into prompts and models, establish tracing and telemetry, and prepare incident response playbooks.
• Partnering & Enablement: Co-develop solution documents with Solution Architects, create internal reference implementations, templates, and standards. Mentor and guide implementation teams.

Required Qualifications

• 8+ years of experience in AI/ML and data engineering, with at least 3 years architecting production AI systems at enterprise scale.
• Demonstrated success delivering LLM-powered applications (chatbots, RAG, agents) and classical ML solutions (forecasting, classification, ranking) in production environments.
• Deep hands-on AWS experience, including services like Bedrock, SageMaker, S3, Lake Formation, Glue, Redshift/Athena, Lambda, EKS/ECS, VPC, IAM, and KMS.
• Proficiency with vector search and retrieval systems, embeddings, re-ranking, and filtering—experience with at least one managed vector solution (Kendra, OpenSearch, MongoDB Atlas Vector, Pinecone).
• Strong understanding of MLOps best practices: CI/CD pipelines, model registries, versioning, rollbacks, and observability.
• Solid grasp of security and compliance: IAM, encryption, network isolation, tokenization, and regulated data (PHI/PII) handling.
• Proven ability to model and optimize for cost and performance (GPU sizing, autoscaling, caching, quantization).
• Excellent communication, documentation, and collaboration skills across technical and business stakeholders.

Preferred / Nice to Have

• Experience with Azure AI (Azure OpenAI, Cognitive Search, Synapse) for multi-cloud or hybrid deployments.
• Familiarity with Snowflake Cortex/ICEBERG or Databricks MosaicML/Dolly.
• Knowledge of agent frameworks (LangChain, LlamaIndex) and their deployment on AWS.
• Hands-on experience with Kafka/MSK and real-time streaming architectures.
• FinOps mindset and prior management of large-scale compute or token-based budgets.
• Background in regulated industries such as healthcare or financial services; FedRAMP experience is a plus.
• Cloud or AI certifications (e.g., AWS ML Specialty, AWS Solutions Architect Pro, Azure Data/AI Engineer).

Job Type: Contract

Application Question(s):

• What is your Work Authorization Status?
• Do you have experience with Azure AI? if yes then how many years?
• what Cloud or AI certifications you have?
• How many total years of experience you have as an AI Architect?

Work Location: Remote