Corbin Floyd

Applied AI and business analytics candidate.

I look for where AI can fit into a real workflow, test it against examples, and write the handoff notes another person would need to trust it.

View TradingAgents work example View AI Fluency evidence
TradingAgents research system framework diagram with graph-memory snapshot counts.

TradingAgents framework view: 973 nodes, 2,678 edges, 265 episodes.

AI evaluation

Model-output review, multimodal review, rubric scoring, instruction-following checks, and unsupported-claim review.

High-trust support

Fraud, identity verification, account access, suspicious activity, escalation, and policy-heavy documentation.

TradingAgents proof

A personal project example for multi-step agent work, graph-memory inspection, output review, and handoff notes.

AI Fluency 4D work

Skill Finder evidence mapped to Delegation, Description, Discernment, Diligence, and approval boundaries.

Main Work Example

TradingAgents personal project example

A finance-adjacent AI workflow I use to practice the same pattern a real organization would need: understand the workflow, inspect the agent outputs, check the limits, and document what another person can trust.

Personal project example. No investment advice. No claim of trading performance.

  • Project Overview image and framework view are prepared for public use.
  • Graph-density still is redacted so node labels are not readable.
  • Full proof video stays private until a redacted clip or controlled-sharing decision exists.
Open the case page
Redacted dense TradingAgents graph structure with labels blurred.
973nodes
2,678edges
265episodes

Secondary Evidence

AI Fluency: Skill Finder evidence

Skill Finder is a personal AI fluency project that made my capability-search workflow more accountable. I designed a process that searches existing options, inspects important files, scores candidates, records risks, and asks for approval before installation.

Project evidence only. Certificate files are required before any certificate claim.

Delegation

Define the problem, understand the platform, and choose what an assistant should handle.

Description

Specify the output, process, and performance bar before relying on a result.

Discernment

Check relevance, currentness, risks, dependency readiness, and fallback behavior.

Diligence

Record what changed, keep approvals visible, and separate local rules from global ones.