Role: AI Engineer – Agentic AI, Commercial Banking
Location: Toronto, hybrid 3 days/week (Tue/Wed anchor)
Contract – 6 months + 1-year extension
FTE conversion possible after ~2 years
Top Must Haves:
- 2+ years building agentic AI shipped to production (not POCs)
- Agent evaluation – offline/online evals, golden datasets, LLM-as-judge, regression gates in CI/CD
- Strong Python engineering (FastAPI, Pydantic, pytest, async, APIs)
- Applied data science foundation (pandas, NumPy, scikit-learn, SQL, experimentation, error analysis)
- RAG and context engineering (hybrid search, reranking, grounding, token budgets)
Required
- 2+ years relevant experience with applied DS/ML foundation and hands-on production retrieval, GenAI or agentic AI delivery
- Strong Python engineering (packaging, typing, testing, async, APIs, logging)
- Applied data science: pandas, NumPy, scikit-learn, SQL, statistics, experimentation, error analysis
- LLM APIs, prompt engineering, structured outputs, tool calling, RAG or workflow orchestration
- Agent architecture: tools, state, memory, routing, planning, retries, fallbacks, HITL
- Evaluation dataset design and automated eval pipelines (deterministic checks, rubrics, human eval, calibrated LLM-as-judge)
- Enterprise RAG: embeddings, BM25/hybrid, reranking, retrieval-quality evaluation
- Git, CI/CD, Docker, cloud deployment, monitoring
- Understanding of AI risk and controls
- Degree in CS, DS, engineering, stats, math or equivalent experience
Preferred
- LangGraph, MCP or comparable orchestration
- PyTorch/TensorFlow, gradient boosting, NLP, anomaly detection
- LangSmith, Arize Phoenix, MLflow, Azure AI Foundry, Bedrock evals
- AWS or Azure AI services, vector search, model gateways
- Financial services / Commercial Banking (lending, underwriting, credit, onboarding, RM, treasury, payments)
- Regulated-environment production-readiness reviews
- Elasticsearch/OpenSearch, Azure AI Search, pgvector; LangChain, LlamaIndex, DeepEval, TruLens
Engagement notes (from Sundar, Oct 30)
- Must have 2+ yrs agentic AI shipped to production, not POCs. Banking domain (commercial, retail, treasury, payments, pricing) a plus
- FTE conversion possible after ~2 years
About the role
Turn business requirements and solution designs into production-grade Python agents, data-grounded AI capabilities, evaluation pipelines, and observable AI services for Commercial Banking copilots (RM prep, credit analysis, underwriting, pricing). Works with the AI Solution Architect and AI Product Manager and owns the tuning loop from data and context preparation through offline evaluation, online monitoring, user feedback, and continuous improvement.
What you will do
- Build production generative/agentic AI apps in Python: agents, tools, workflow graphs, state, memory, APIs, background tasks, human-in-the-loop
- Implement approved patterns for orchestration, model/tool integration, data access, security, resiliency, observability
- Convert user journeys and acceptance criteria into incremental deliverables and measurable AI behaviours
