About the role
JD — AI Solution Architect (Fintech · Web, Mobile & Agentic)
Role
The AI Solution Architect turns business goals into secure, scalable AI/ML and agentic solutions: reference architecture, integration, and governance—with payments-aware risk discipline.
Experience: 12–15 years in software/solution architecture; strong recent exposure to GenAI, agents, and production integrations.
Responsibilities
Area
Expectations
AI & agentic design
End-to-end solutions: LLM + tools (e.g. MCP-style),RAG/memory where grounding helps, guardrails, durable workflows (state, HITL, idempotency for multi-step flows).
Strategy & stack
Prioritize high-value use cases; define cloud/hybrid targets (AWS and/orAzure as applicable),model-agnostic patterns, cost/latency trade-offs.
Integration
APIs and events for web, mobile, and agentic clients;goal-oriented machine-facing contracts alongside traditional channels; integration withCRM (e.g. Salesforce) and core platforms.
Security & access
AuthN/AuthZ for users and automated actors (agents, services): least privilege, audit, secrets,no LLM→database direct access—tools/APIs only.
Operations
Observability for quality, latency, cost, abuse; evaluation and improvement loops (not one-off launches).
Compliance & ethics
Secure development, data minimization, alignment withfintech/regulatory expectations for AI use.
Evolution
Monolith → modular/API-first roadmaps (strangler, bounded contexts) includingagent-ready surfaces.
Technical profile
- AI/ML & agents: Practical depth with examples such as PyTorch / TensorFlow (classical ML), LangChain / LangGraph / LlamaIndex or equivalent (orchestration & RAG), MLflow or cloud ML ops where used.
- Cloud: AWS (primary ask) and/or Azure; identity, networking, observability.
- Data: Yes — required: pipelines (batch/stream), quality, and timely data for models/agents—not optional at architect level.
Integration: REST/events, API gateways; Salesforce and enterprise systems in target-state designs alongside agents.