async function validateSystem() { const checks = await runDiagnostics(); if (checks.every(c => c.passed)) { return deployAgent(); } } class AgentWorkflow { constructor(private readonly evaluator: Evaluator) {} async optimize(prompt: string): Promise<Result> { let best = await this.evaluator.baseline(prompt); for (let i = 0; i < MAX_ITERATIONS; i++) { const candidate = await this.mutate(best); if (await this.evaluator.score(candidate) > best.score) { best = candidate; } } return best; } } interface SafetyCheck { id: string; severity: 'critical' | 'warning' | 'info'; validate: () => Promise<boolean>; } const RISK_LAYERS = [ 'organizational_culture', 'supervision_oversight', 'preconditions_unsafe_acts', 'unsafe_acts_errors', ]; async function validateSystem() { const checks = await runDiagnostics(); if (checks.every(c => c.passed)) { return deployAgent(); } } class AgentWorkflow { constructor(private readonly evaluator: Evaluator) {} async optimize(prompt: string): Promise<Result> { let best = await this.evaluator.baseline(prompt); for (let i = 0; i < MAX_ITERATIONS; i++) { const candidate = await this.mutate(best); if (await this.evaluator.score(candidate) > best.score) { best = candidate; } } return best; } } interface SafetyCheck { id: string; severity: 'critical' | 'warning' | 'info'; validate: () => Promise<boolean>; } const RISK_LAYERS = [ 'organizational_culture', 'supervision_oversight', 'preconditions_unsafe_acts', 'unsafe_acts_errors', ];

Building Systems That Survive Human Nature.

Abraham Haskins, PhD

Senior Systems Architect & Agentic Coding Specialist

→ Request Technical Walkthrough↓ Download Sanitized CV
↓
PythonRC#C++TypeScriptAgentic WorkflowsOPro OptimizationBayesian NetworksDockerUnityFigmaISO 9001LLM TuningHuman FactorsPythonRC#C++TypeScriptAgentic WorkflowsOPro OptimizationBayesian NetworksDockerUnityFigmaISO 9001LLM TuningHuman Factors

// Selected Works

Safety-Critical Root Cause Analysis

Investigating high-visibility structural non-conformance in aerospace production.

Domain

Human Factors Forensics, Systemic Risk Mitigation

Outcome

Delivered dual-phase remediation: Immediate containment protocols + fundamental logic redesign of installation architecture.

Request Discussion
UserAgentEvalOut

Automated LLM Tuning Loop

Scaling AI adoption across a 30,000-person engineering organization.

Domain

Agentic Coding, Large Language Model (LLM) Optimization

Outcome

Engineered self-healing OPro loops to iteratively optimize prompt performance without human intervention.

Request Discussion

Bayesian Defect Prediction

Quantifying uncertainty in complex manufacturing environments.

Domain

Probabilistic Modeling, Data Science

Outcome

Built the General Inspection-Error Model to predict quality escapes based on environmental variables.

Request Discussion
about.sh

$ whoami

Transitioning from Ph.D. research in Human-AI Teaming to building production-grade autonomous systems.

$ cat specialty.txt

Specialist in Agentic Coding—moving beyond scripts to autonomous, decision-making agents. I do not just analyze human error; I architect the automated systems that prevent it.