I build high-performance frontend systems and AI-driven applications for financial services — specializing in data-intensive React and TypeScript interfaces where performance, correctness, and usability all matter.
My background spans equities trading and banking platforms at firms including Nomura and JPMorgan, where I built real-time blotters, dashboards, and trade booking systems under continuous data load. I understand what financial UIs need to do and why — order lifecycle, streaming state, audit trails, and the domain logic underneath.
Recently I've extended into AI integration — building RAG pipelines, LLM-powered workflows, and AI-assisted interfaces that bring intelligent features into production systems. I apply the same engineering discipline to AI that I bring to frontend work — clear boundaries, validation, and observable behavior.
I actively use AI-assisted development workflows, applying the same engineering rigor to AI coding tools as I do to production systems. Structured prompting, validating generated code, and integrating AI output with proper testing and review — accelerating delivery without sacrificing quality or maintainability.
Core expertise
React · TypeScript · WebSockets · Real-time data · Financial systems · RAG pipelines · LLM integration · Python · FastAPI · PostgreSQL · AWS · GCP
Featured Systems
FlowDesk
GitHubReal-time order management system for trading teams — live blotter, P&L tracking, full audit trail, and AI-powered summarization and NLP filtering.
Architecture: Event-driven • WebSocket streaming • Component-based state management
Production Use Cases: Equities: Order lifecycle management, real-time P&L, risk exposure tracking • AI: Natural language order filtering, session summarization, EOD report generation • Audit: Immutable field-level change trail with actor and timestamp
Ask my Docs (RAG)
GitHubRetrieval-augmented Q&A over documents and data — ingestion and chunking, embedding pipelines, vector retrieval with reranking, grounded answers with citations, and evaluation hooks for production rollout.
Architecture: Event-driven ingestion • Embedding + vector index • Retriever + reranker • LLM orchestration • Observability and eval
Production Use Cases: Knowledge bases: Internal docs and policy Q&A • Support: Ticket and runbook search with sources • Compliance: Cited answers and audit-friendly retrieval logs
Data Mind
GitHubNatural-language analytics and exploration layer — connect sources, ask questions in plain language, surface metrics and relationships, and trace results back to underlying tables and definitions.
Architecture: API-backed services • Structured + semantic query paths • Metric and entity models • Lineage and audit-friendly responses
Production Use Cases: Analytics: Ad hoc NL queries over approved datasets • Self-serve: Consistent metrics and definitions for teams • Governance: Actor, timestamp, and source attribution on answers
Other Systems
AI Governance Platform
Enterprise AI control plane with pluggable policy enforcement, dual-checkpoint validation, full audit trails, and human-in-the-loop workflows — built for regulated environments that need auditability and controlled execution.
Agentic Enterprise Agent
Constrained planner–executor for enterprise: the LLM proposes a plan (tool calls); validation and execution are deterministic and auditable. Clear audit trail, no black-box runs, governance plug-in before any tool executes.
Technical Work and Insights
ADRs
├─ Policy Outcome Model Design
Framework for modeling and evaluating policy outcomes in AI governance systems.
├─ Sync vs Async Communication Patterns
Decision framework for choosing synchronous vs asynchronous communication in distributed AI systems.
└─ Modular Monolith to Microservices Strategy
Migration approach for evolving monolithic AI platforms into distributed microservices architecture.
Technical Writing (coming soon)
└─ How the Governance Platform scales 10x → 1000x
Documenting the evolution and scaling challenges of AI governance platform.
Case Studies
├─ NLP-to-SQL Interface for Enterprise Databases
Enables non-technical users to query enterprise databases in plain English with validation and visualization, reducing analysis time from hours to minutes.
└─ RAG-Powered Document Intelligence System
Production RAG system providing instant, context-aware access to enterprise documentation with semantic search and source attribution.
Contact
Memphis, TN | Previously New York, NY