Dmitri Farafonov

Set cloud-native and AI engineering standards at a regulated fintech and bank, then built the platforms that put them into practice.

• AI strategy and governance: wrote the company's AI agent standard (patterns, guardrails, integration rules) and built and run a fleet of autonomous agents for ticket triage, code review and domain knowledge across multiple product platforms.
• AI knowledge base and legacy code analysis: designed and built a federated, RAG-based engineering knowledge base: AI agents analyze source code and docs to write and refresh a central hub and a wiki per platform, and agents and engineers query it through an MCP search service, so AI answers rest on curated, cited knowledge.
• AI-assisted engineering at scale: built the developer CLI and shared AI skills library that let teams use AI coding assistants day to day, and helped engineers across platforms write and share their own reusable skills, achieving up to 10x performance improvements.
• Wrote and own the company's Kubernetes GitOps standard (Helm + ArgoCD), approved by the Enterprise Architecture Review Board, and built the deployment toolchain behind it across multiple product platforms.
• Defined the standard for automated container validation (post-change checks, daily regression, post-deploy verification, production health checks); wrote the architecture for secure ingress, egress and multi-tenancy on AKS; introduced Azure Data API Builder as the standard GraphQL data-access layer.
• Leads hands-on: 700+ pull requests and 3,200+ commits across 140+ repositories, plus 600+ code reviews for 80 engineers over two years.

Member of the Distinguished Engineers program, accountable for technical direction across a 1,400-developer organization. Led 25 engineers (6 direct reports, 4 pods) and owned future tech stacks, shared frameworks and engineering standards.

• Worked across the Optis card issuing platform: authorization, clearing and settlement, account management, billing, rewards, disputes, fraud and collections.
• Delivered AI-driven products for real-time fraud detection, automated disputes workflows and real-time collections decisioning.
• Architected high-volume processing systems handling 15,000 financial transactions per second, and was lead architect for 800+ RESTful APIs handling 3,500 calls per second.
• Shipped one-time-passcode notifications for Apple Pay and Google Wallet, and USPS-integrated mail tracking for cards and statements.
• Led engineering-culture transformation: firm-wide CI/CD for 530 repositories with security scanning, enforced test coverage and segregation-of-duties controls, plus TDD and Clean Code practice, tracked through measurable 1-, 3- and 5-year goals and Splunk dashboards.
• Led modernization of 30+ legacy applications into cloud-native microservices across OpenShift (Azure and on-prem), Azure AKS and VMware/Pivotal; managed strategic vendor relationships and negotiations, saving millions of dollars.

• Took over check duplicate detection after multiple failed attempts and delivered it as a standalone API: 20M checks per 3-hour batch window compared against 13B+ historical records in under 1 ms.
• Designed a custom in-memory engine on Bloom filters (0.6 ms average per call, 4 µs per comparison), run Active/Active across data centers; it ran 18 months without a restart.
• Rewrote demand deposit account (DDA) systems to sustain 6,000+ financial transactions per second.
• Tuned JEE applications on WebSphere for z/OS (CPU and thread profiling, JPA optimization for DB2 on z/OS) and secured them with centralized SSL certificate management and Jasypt/Bouncy Castle encryption.