Solutions Architect - AI

Key Responsibilities

• Define and own the enterprise AI architecture for retail use cases, aligning with business priorities and technology strategy.
• Develop reference architectures, patterns, and standards for AI/ML and Generative AI solutions, with an emphasis on open-source-first design principles.
• Translate retail business problems — across merchandising, supply chain, stores, marketing, and e-commerce — into scalable AI solution blueprints.
• Partner with business and product leaders to identify and prioritize high-impact AI opportunities.
• Champion open-source AI frameworks and tooling (e.g., Hugging Face, LangChain, LlamaIndex, Ray, MLflow, Feast) as the default approach before evaluating commercial alternatives.

Open-Source AI Architecture

• Lead the selection, evaluation, and integration of open-source AI and ML frameworks into Company’s enterprise architecture.
• Design reusable patterns for open-source LLM deployment, fine-tuning, and serving (e.g., vLLM, Ollama, llama.cpp, OpenLLM).
• Establish governance standards for open-source model usage, including licensing review, security scanning, and model provenance tracking.
• Build internal capability around open-source foundations to reduce vendor lock-in and accelerate experimentation velocity.
• Evaluate and adopt emerging open-source agentic frameworks (e.g., AutoGen, CrewAI, LangGraph) for retail automation use cases.

AI Solutioning & Design

• Architect end-to-end AI solutions, including data ingestion, feature engineering, model training, inference, and system integration.
• Design AI systems for core retail domains such as: Search, recommendations, and personalization Demand forecasting, inventory optimization, replenishment, and allocation Pricing and markdown optimization AI assistants and copilots for store, merchandising, and supply-chain teams
• Define integration patterns between AI services and retail platforms (POS, OMS, WMS, CRM, e-commerce).
• Lead architectural reviews, ensuring solutions meet performance, scalability, security, cost, and reliability requirements.

AI Observability

• Define and implement an AI observability framework covering model performance monitoring, data drift detection, prediction quality tracking, and system health across all production AI systems.
• Establish real-time and batch monitoring pipelines for model inference using tools such as Evidently AI, Arize, WhyLogs, Fiddler, or equivalent open-source platforms.
• Design standardized dashboards and alerting for model degradation, data skew, latency SLO breaches, and feature store anomalies.
• Build feedback loop infrastructure to capture ground-truth labels and enable continuous model evaluation in production.
• Define observability standards for GenAI and LLM systems, including hallucination rate tracking, prompt/response logging, latency percentiles, and cost-per-query attribution.
• Partner with MLOps and Platform Engineering to embed observability as a first-class requirement in every AI system from Day 1.

AI Security

• Serve as the AI security authority for Company, owning the threat model for all AI and ML systems in production.
• Define and enforce secure-by-design standards for model development, training data handling, inference APIs, and GenAI integrations.
• Architect defenses against AI-specific attack vectors, including prompt injection, model inversion, adversarial inputs, data poisoning, and supply chain risks in open-source model adoption.
• Establish data privacy controls for AI pipelines, ensuring compliance with applicable regulations (e.g., CCPA) and internal data governance policies.
• Lead AI red-teaming and adversarial testing exercises to proactively identify and remediate security gaps before production deployment.
• Partner with Information Security, Legal, and Enterprise Risk to maintain an AI risk register and align AI security posture with the organization’s broader cybersecurity framework.
• Define guardrails, content filtering, and human-in-the-loop safeguards for all customer-facing and associate-facing GenAI applications.

MLOps, GenAI & Governance

• Establish MLOps and AIOps practices, including CI/CD for models, automated retraining, monitoring, drift detection, and cost controls.
• Define standards for Generative AI and LLM usage, including multi-RAG architectures, MCP, and vector search.
• Define prompt orchestration, tool-calling, and agentic workflow patterns.
• Ensure AI solutions comply with data privacy, security, and responsible AI principles.
• Partner with Security, Legal, and Enterprise Architecture to align AI solutions with governance and risk standards.

AI Productivity Tooling

• Mandate and model the daily use of AI-native productivity tools across all architecture and delivery work.
• Evaluate, recommend, and govern the enterprise use of tools including: Microsoft Copilot, Cursor, Glean, Claude, or equivalent or emerging AI productivity platforms as the market evolves.
• Define standards and guardrails for enterprise AI tool adoption, including data classification policies governing what information may be shared with each platform.
• Train and upskill engineering and cross-functional teams on effective use of AI productivity tooling to multiply output and reduce time-to-delivery.

Technology Evaluation, Implementation & Delivery

• Work closely with AI Engineers, ML Engineers, Data Engineers, and platform teams to ensure architectures are production-ready and executable.
• Provide hands-on guidance during implementation, including reference code, pipelines, schemas, and infrastructure patterns.
• Evaluate and recommend AI SaaS solutions, cloud services, and frameworks (AWS, Azure, GCP, Databricks, Snowflake, etc.).
• Lead build vs. buy vs. open-source decisions and support vendor selection for AI capabilities.