AI Solution Architect

STACK Infrastructure is an award-winning industry leader in building, owning, and operating highly efficient, cost-effective wholesale, colocation, and cloud data centers. They are seeking an AI Solution Architect to serve as the senior technical authority and delivery lead for enterprise AI solutions, owning the design, engineering, and deployment of AI systems while driving technical adoption across the organization.

Responsibilities:

• Design, develop, and deploy production GenAI solutions including custom assistants, multi-agent systems, and LLM-powered workflow automation — with hands-on ownership of every layer from prompt design through inference endpoint
• Architect and implement smart LLM routing logic — designing multi-model routing systems that dynamically select the right model based on query complexity, cost thresholds, latency requirements, and data residency constraints. Implement fallback chains, load balancing, and model arbitration patterns for production reliability
• Build and optimize Retrieval-Augmented Generation (RAG) pipelines end to end — including document ingestion strategy, chunking and overlap logic, embedding model selection and tuning, vector store architecture, hybrid retrieval design, reranking layers, and context window management for enterprise knowledge applications
• Engineer and maintain prompt libraries, system prompts, chain-of-thought patterns, and prompt chaining logic for all production LLM applications — maintaining a versioned prompt registry in GitLab with structured testing and evaluation before promotion
• Build LLM fine-tuning and alignment pipelines on Databricks MLflow — including supervised fine-tuning (SFT), parameter-efficient fine-tuning (LoRA, QLoRA), and RLHF
• Implement guardrails, content filtering, hallucination detection, and output validation layers to ensure AI responses meet safety, accuracy, and compliance standards established by the governance framework
• Implement and deploy predictive AI solutions that surface forecasts and recommendations to business users — building on feature specifications and model evaluation criteria provided by the AI Data Scientist
• Engineer the prescriptive AI layer — translating model outputs and recommendations into automated decisions, workflow triggers, and user-facing actions within business systems
• Build model serving infrastructure: inference endpoints, prediction APIs, caching layers, and fallback logic that ensure production models meet latency, reliability, and cost requirements
• Integrate AI model outputs into operational workflows, automated alerts, and decision support tools so insights reach end users in context and at the right moment
• Design and build the full agentic AI architecture for Phase III of the enterprise AI roadmap — including agent orchestration layers, tool registries, memory systems (short-term, long-term, episodic), state management, and inter-agent communication protocols
• Architect multi-agent workflows that span enterprise systems — designing how agents interact with NetSuite, Procore, SharePoint, Workday, and operational platforms to complete complex multi-step business processes autonomously
• Build and maintain the agent tool library — the set of callable Python functions, REST API wrappers, and enterprise connectors that agentic systems invoke to read data, trigger actions, and write outputs to business systems
• Implement intelligent routing within agentic systems — designing intent classification layers, task routing logic, and dynamic tool selection patterns that direct agent actions to the right model, tool, or human escalation path based on query type and confidence
• Own the full agentic AI lifecycle from design through production scaling — this is not a handoff function, it is an end-to-end engineering ownership responsibility
• Own the AI intelligence layer on Databricks — building, maintaining, and iterating all AI and ML workloads including model training notebooks, inference pipelines, feature transformation jobs, embedding generation logic, and LLM orchestration workflows running on Databricks ML Runtime
• Write production-grade Python code across all AI solution engineering work — including Azure Functions for serverless AI triggers, custom API wrappers, LLM chain implementations, agent tool functions, data transformation scripts, and automation pipelines connecting AI systems to enterprise platforms
• Build and maintain Databricks Jobs and Workflows orchestrating multi-step AI pipelines — coordinating data preparation, model inference, output transformation, prompt execution, and downstream delivery in a single governed execution context
• Leverage Databricks Mosaic AI capabilities — including AI Playground, Model Serving, Vector Search, and LakeFlow Connect
• Optimize Databricks AI workload performance — tuning notebook compute, cluster sizing for training jobs, and serving endpoint configuration — in coordination with the Sr. Data Platform Engineer who owns cluster policy and platform infrastructure
• Own and maintain the GitLab repository structure for all AI initiatives — following the initiative-based repo architecture under the dedicated AI workspace, with consistent branching conventions, commit standards, and merge request workflows across all solution engineering work
• Build and maintain CI/CD pipelines in GitLab for AI solution deployment — automating testing, validation, model promotion, and environment deployment for GenAI applications, inference APIs, Azure Functions, and Databricks Jobs
• Maintain GitLab as the single source of truth for all AI solution code, configuration, model artifacts, and deployment history — ensuring every production AI system has a complete, auditable engineering record
• Build and maintain integrations between AI systems and enterprise platforms including NetSuite ERP, Procore, Asana, Microsoft 365, and Workday — enabling AI-powered insights, automation triggers, and real-time data flows
• Design and engineer the API and connector layer that AI agents and applications use to read from and write to enterprise systems — owning authentication (OAuth 2.0, API keys, managed identities), authorization, error handling, retry logic, and rate limiting
• Build data ingestion scripts and connectors in Python for AI-specific data feeds — pulling targeted data from enterprise sources (NetSuite reports, Procore project data, SharePoint documents, Teams activity) into AI-ready formats for model consumption
• Design and implement Azure Function-based ingestion triggers — event-driven Python functions that capture real-time data updates from enterprise systems and route them to AI processing pipelines without requiring full batch pipeline infrastructure
• Establish and enforce engineering standards for AI solution development, testing, deployment, monitoring, and deprecation — covering Python code quality, GitLab workflow discipline, Databricks notebook standards, prompt versioning, and API documentation
• Own technical solution monitoring: system uptime, API latency, error rates, token consumption, Databricks job health, and infrastructure reliability for all deployed AI solutions. Coordinate with the AI Data Scientist on model performance signals that indicate architectural review is needed
• Ensure all AI solutions comply with security, privacy, data governance, and regulatory requirements in collaboration with IT, legal, and cybersecurity teams