Principal AI Solutions Engineer

Scotiabank • Toronto, ON

21h • Onsite

About The Position

The Principal AI Solutions Engineer is the function-wide technical authority for how GenAI and agentic AI capabilities from the enterprise AI Platform are translated into production-grade solutions across Scotiabank’s business lines — retail banking, commercial banking, wealth management, and enterprise shared services. They own end-to-end solution architecture for the most complex, high-value AI engagements, authoring reference solution patterns, integration blueprints, and adoption playbooks that squads and business unit teams implement when deploying capabilities from the GCP AI Platform (Vertex AI, Vertex AI Agent Engine, Apigee X LLM Router, MCP Control Gateway, Model Armor) and the Azure Databricks ML Platform. This role engages as a technical peer with Principal AI and ML Engineers, business unit CTOs, and the Chief AI Officer’s office — shaping how AI platform capabilities are surfaced, governed, and scaled across the enterprise — and is accountable for technical quality across the most strategically significant AI solution programmes in AI and Agentic Engineering.

Requirements

• 10+ years of software engineering, solutions architecture, or AI/ML engineering experience, with at least 5 years leading solution architecture for production AI systems at enterprise scale — owning technical outcomes across multiple business lines or customer segments, not internal platform components only.
• A demonstrable track record of translating ambiguous, high-stakes business problems into production AI systems — reference solution patterns, integration blueprints, or adoption playbooks adopted and implemented by engineering teams across 30+ engineers or multiple business units, with demonstrated business outcomes.
• Mastery of the enterprise GenAI and agentic AI solution stack: multi-agent system design on Vertex AI Agent Engine and Google ADK / LangGraph (including the deep agents harness), RAG architecture spanning Vertex AI Vector Search and Mosaic AI Vector Search, LLM integration through Apigee X and Azure APIM, MCP tool ecosystem design via the MCP Control Gateway, and the three-pillar agent model (dev-built, business-user-built via Agentspace and Genie, agent-built) — at the depth required to resolve solution design ambiguity without escalation.
• Proven ability to engage at VP and C-suite level: framing technical solution trade-offs in business value terms, presenting architecture positions under regulatory scrutiny, and influencing investment decisions by translating aggregate solution demand into platform roadmap requirements.
• Deep fluency in the regulatory constraints that shape AI solution design in a federally regulated Canadian financial institution: OSFI E-23 model risk management requirements for model validation, approval, and monitoring; OSFI B-13 technology and cyber risk implications for agentic systems; OSFI B-10 third-party risk controls at the LLM vendor interface; PIPEDA, AIDA, and Quebec Law 25 data residency and consent obligations — and the practical experience embedding these as solution-layer architecture controls, not documentation-only compliance artefacts.
• Track record of designing and executing AI solution evaluation programmes: LLM evaluation harness design using Mosaic AI Agent Evaluation and Vertex AI Evaluation, responsible AI red-teaming methodology aligned to OWASP LLM Top 10, evaluation metrics frameworks that address business KPIs alongside technical quality, and the structured evidence packaging needed for OSFI E-23 pre-production approval.
• Experience leading technical discovery and feasibility assessment for AI programmes at enterprise scale: data dependency analysis, platform integration complexity assessment, regulatory risk scoping, and the ability to produce a structured go/no-go recommendation that VP-level stakeholders can act on.
• Exceptional technical communication: solution architecture documents precise enough for engineers to implement and clear enough for business unit executives to fund, and the ability to translate failure modes and regulatory risk into language that resonates with non-technical governance forums.
• External technical presence in enterprise AI, financial services AI, or responsible AI: conference presentations, standards contributions, or published solution methodology that establishes credibility beyond the organization.

Responsibilities

• Own end-to-end solution architecture for the highest-complexity, highest-business-value AI engagements across Scotiabank’s business lines — including multi-agent agentic solutions built on Vertex AI Agent Engine and Google ADK / LangGraph, RAG solutions spanning Vertex AI Vector Search and Mosaic AI Vector Search, LLM-powered automation integrated through the Apigee X LLM Router, and ML-assisted decision systems on the Azure Databricks ML Platform — from discovery through to production deployment inside the Montreal walled garden.
• Author the function-wide AI solutions reference library: canonical solution patterns, integration blueprints, and architectural decision records for recurring AI use-case categories — conversational AI, document intelligence, agentic workflow automation, real-time recommendation, risk signal generation — written at the precision required for squads to implement without ambiguity and at the clarity required for business unit CTOs to approve.
• Lead technical discovery and solution design for strategic AI programmes — partnering with business unit product owners and domain experts to translate ambiguous business problems into structured AI solution designs with explicit feasibility, regulatory risk, data dependency, and platform integration assessments before a squad commits to delivery.
• Define and own the AI solution adoption playbook for business line teams: onboarding architecture for Vertex AI Agent Engine and Agentspace, integration patterns for the MCP Control Gateway’s tool ecosystem, data access and data sharing patterns compliant with PIPEDA and Quebec Law 25, model selection guidance for Gemini, Claude, GPT-5.x (via APIM), and open-weights (Llama), and the solution-tier controls framework that separates dev-built, business-user-built, and agent-built AI in production.
• Engage as a technical peer with Principal AI and ML Engineers, Enterprise Architecture, and the Chief AI Officer’s office in cross-function solution governance forums — providing authoritative technical positions on how platform capabilities are exposed to business lines, influencing the AI Platform roadmap based on aggregate solution demand signals, and translating OSFI E-23, OSFI B-13, OSFI B-10, PIPEDA, AIDA, and Quebec Law 25 requirements into solution-layer controls embedded in adoption playbooks.
• Lead proof-of-concept and technical validation programmes for strategic AI use cases — designing evaluation methodology, authoring evaluation frameworks using Mosaic AI Agent Evaluation and Vertex AI Evaluation, interpreting results with business stakeholders, and making the go/no-go production recommendation with a concrete technical risk register.
• Own function-wide AI solution quality standards: define the solution-layer quality bar that sits above squad-level engineering standards — minimum evaluation coverage, responsible AI instrumentation via Model Armor and Sensitive Data Protection, Azure AI Content Safety + Prompt Shields, explainability requirements for model-assisted decisions in regulated product lines, and the audit evidence packaging required for OSFI E-23 model risk management review.
• Identify systemic solution delivery friction — integration complexity, data access blockers, regulatory ambiguity, platform capability gaps — and drive resolution through platform roadmap influence, architecture standard updates, or direct co-design with the AI Platform engineering team; produce the gap analysis that informs the function’s quarterly platform investment decisions.
• Mentor Senior and Staff AI Solutions Engineers and cross-functional embedded technical leads: conduct solution architecture reviews, develop engineers toward independent solution leadership on complex engagements, and raise the function-wide AI solution delivery capability.
• Represent AI and Agentic Engineering’s solution delivery capability externally: present at industry forums on enterprise AI adoption patterns, contribute to responsible AI and financial services AI standards bodies, and build Scotiabank’s credibility as a reference architecture in enterprise AI at scale.

Benefits

• Upskilling through online courses, cross-functional development opportunities, and tuition assistance.
• Competitive Rewards program including bonus, flexible vacation, personal, sick days and benefits will start on day one.
• Free tea & coffee, universal washrooms, and lots of space for team collaboration.
• Opportunities for community engagement & belonging with our various programs.

What This Job Offers

• Job Type: Full-time
• Career Level: Principal