VARITE INC · 11 hours ago
Solution Architect – Agentic Commerce & Payments
VARITE INC is an award-winning minority business enterprise providing global consulting and staffing services. They are seeking a qualified Solution Architect to lead the design of a 'Zero-Exit' fulfillment engine for a large retailer, focusing on Agentic AI orchestration and E-commerce Payments infrastructure.
Information Technology & Services
Responsibilities
Design the decision engine that routes requests between Deterministic Workflows (API-based, low risk) and Agentic AI Workflows (Headless browser-based, high complexity)
Architect the 'Headless Browser' infrastructure (using tools like Playwright/Puppeteer) to navigate non-standard third-party checkout flows without user intervention
Define the technical flow for the 'Double-Ledger' system: bridging the customer’s native payment with the backend generation of Virtual Commercial Cards (VCC) for third-party settlement
Ensure the architecture is PCI-DSS Compliant, ensuring customer raw payment data never leaves the secure vault/enclave
Implement Enterprise Guardrails using RAG (Retrieval-Augmented Generation) to prevent the Agent from purchasing wrong items or hallucinating prices
Design the 'Human-in-the-Loop' escalation path for when the Agent encounters soft blocks (Captchas, MFAs) or high-risk transaction anomalies
Map the technical workflow for Return Orchestration: How the system autonomously retrieves return labels from a third party and presents them to the customer natively
Qualification
Required
Agentic AI Experience: Proven experience deploying LLM-based agents that use tools (Function Calling) to interact with the web (e.g., LangChain, AutoGPT concepts)
Browser Automation at Scale: Deep expertise in Playwright, Puppeteer, or Selenium for handling dynamic DOMs, anti-bot detection, and session management
Fintech / Payments Depth: Experience with Virtual Card Issuing APIs (e.g., Stripe Issuing, Marqeta, Adyen) and understanding of 'Merchant of Record' liabilities
Security First: Demonstrated history of building architectures that separate PII/PCI data from the AI inference layer