Selected work

Case study · 2024 — Present

Managing

Staff AI Engineer–oriented trading platform: agents (Semantic Kernel, MCP, Hermes, n8n), Optuna research pipelines, Spotlight/news/regime context, and distributed live execution.

.NET 10OrleansSemantic KernelMCPHermesn8nOptunaReact 19

The product

The work connects product priorities from trading P&L and operator needs to the systems behind them. Users define their own trading decisions; the platform provides a dependable path from research to paper and live execution.

Platform architecture

The platform uses .NET 10 and Microsoft Orleans for stateful, distributed strategy workloads. React and Next.js support the web experience, while Expo supports mobile monitoring and notifications.

Semantic Kernel, Managing MCP, and the CLI provide controlled tool access for agents. Hermes schedules staged Optuna research pipelines—candidate discovery, optimization, out-of-sample validation, and promotion. n8n routes outcomes, alerts, and human-in-the-loop decisions.

Market context and execution

Spotlight bias and trend analytics, news sentiment, and a Regime Engine provide context for strategies. Recent delivery includes confidence-based position sizing, entry/SL/TP suggestions, trailing stops, multi-timeframe signals, performance analytics, and GMX funding-rate streaming with delta-neutral execution.

My role

As Staff AI Engineer and product owner, I work across product direction, architecture, implementation, and operations. I am extending the platform with Microsoft Agent Framework patterns, RAG, LLMOps governance, AI security controls, and SLO-backed observability.

Evidence placeholder: Add approved scale, reliability, and production metrics before publishing quantitative claims.