Ethan Xiang Claim this profile

Circuit Framework — multi-agent LLM trading research system

A swarm of disposable agents for unattributable web interaction

Policy checks, human approvals, and audit trails for AI workflows.

Curated list for policy-as-code, audit, agents, and LLM safety.

Java toolkit for backward-compatible schema evolution.

Benchmarks for vector retrieval metrics and ranking.

Metrics, tracing, and structured logging primitives.

FastAPI multimodal assistant with moderation and contracts.

Validated model-to-tool dispatch with typed payloads.

Route between local and fallback LLM providers.

Local-first workbench: Rust core, TypeScript UI.

Entity and relation extraction to graph-ready exports.

Infrastructure policy checks with structured findings.

Deterministic feature flags and experiment evaluation.

Go event-driven orchestration with retries and dead-letter queues.

Rust edge proxy with route fallback and timeout handling.

Service catalog and plugin scaffold for developer portals.

Reproducible developer environment setup.

Browser automation with policy hooks for governed agents.

Surface risk hotspots in AI-assisted pull requests.

Short- and long-term memory for agent context.

Streaming chat UI with SSE backend and TypeScript client.

Token cost and latency observability for LLM workloads.

Planning and reconciliation for multi-agent systems.

Rule-based prompt gate for policy enforcement.

Structured filters plus semantic retrieval in one RAG pipeline.

Benchmark autonomous agents against scenario rubrics.

Audit, policy, and risk scoring for AI governance.

Run AI agent task plans with retries and structured output.

Evaluate RAG retrieval quality and answer fidelity.