Founder and Independent AI Researcher

Independent AI research, grounded in engineering experience.

Evening Star AI is Andrew Scott's independent research initiative. His research examines AI security, anomaly intelligence, and governed decisions through technical publications and experimental systems.

AI is only useful when it can be tested, explained, governed, and trusted under pressure.

Andrew Scott

Founder and Independent AI Researcher

Andrew Scott founded Evening Star AI to pursue independent research in AI security and decision systems. His professional background spans more than 15 years across cybersecurity, applied AI, secure software, autonomous systems, and mission-critical engineering.

He founded Evening Star AI around a practical conviction: useful AI has to survive messy data, adversarial behavior, ambiguous requirements, operational constraints, and leadership scrutiny. The research goal is to publish methods and prototype systems that can be tested, challenged, and explained.

Professional background. Andrew's experience combines offensive security judgment, defensive cyber operations, AI/LLM security testing, secure platform engineering, vulnerability intelligence, and safety-critical systems thinking. He has led security assessments of AI-enabled applications, cloud-connected platforms, APIs, retrieval workflows, and agentic systems, with emphasis on prompt injection, insecure retrieval, unsafe tool use, sensitive-data exposure, weak authorization, model misuse, and guardrail failure.

Independent research. Through Evening Star AI / Applied Labs, Andrew builds experimental AI security and decision-support prototypes that turn emerging AI failure modes into concrete workflows, controls, dashboards, and decision support.

Professional experience is presented as background, not as services offered by Evening Star AI. Views expressed are the authors' own and do not represent any employer or affiliated organization. References to professional experience do not imply employer endorsement.

View LinkedIn Discuss the Research

Professional Experience 15+ years across cybersecurity, applied AI, secure software, autonomous systems, defense cyber, and safety-critical engineering.

AI Security LLM security testing, AI assurance, prompt injection, RAG/retrieval risk, unsafe tool use, model misuse, data leakage, guardrails, and NIST AI RMF / OWASP LLM alignment.

Applied AI Builder Python, FastAPI, Docker, local LLMs, structured outputs, retrieval workflows, evaluation harnesses, anomaly detection, drift monitoring, and operator-facing dashboards.

Army Cyber Background Cyber defense leadership with the Michigan Army National Guard's 172nd Cyber Protection Team, including incident-response readiness, threat hunting, vulnerability management, tabletop exercises, and OT/ICS risk.

Safety-Critical Systems Engineering leadership across autonomous, embedded, defense, and cyber-physical systems where reliability, traceability, validation evidence, and risk decisions matter.

Education & Credentials M.S. Electrical & Computer Engineering, MBA, B.S. Electrical Engineering, B.S. Computer Engineering, Trusted AI Safety Expert, GIAC GSLC, GIAC GCPN, GPEN, GISP, CySA+, PenTest+, Cloud+, AWS Cloud Practitioner, CCNA, CEH.

Selected Research Prototypes

Research patterns explored through experimental systems.

Purple Firefish

An AI security gateway concept focused on prompt injection, jailbreak attempts, unsafe tool execution, sensitive-data leakage, policy enforcement, and agentic workflow risk.

Purple Radar

An experimental AI-assisted vulnerability intelligence workflow that turns noisy vulnerability, exploit, and threat context into prioritized findings, risk rationale, and remediation-oriented recommendations.

Candles Edge

An experimental AI analytics platform using Python, FastAPI, Docker, model-driven scoring, anomaly detection, drift monitoring, calibration metrics, historical analogs, structured outputs, explainability summaries, and local LLM support.

Education & Credentials

Technical depth, security discipline, and applied leadership.

Andrew holds an M.S. in Electrical & Computer Engineering with graduate AI/ML research, an MBA, B.S. degrees in Electrical Engineering and Computer Engineering, and a full-stack software engineering certificate. His credentials include Trusted AI Safety Expert, GIAC GSLC, GIAC GCPN, GIAC GPEN, GIAC GISP, CompTIA CySA+, PenTest+, Cloud+, AWS Cloud Practitioner, CCNA, CEH, SAFe Scrum Master, and DFSS Black Belt.

Research Approach

Rigorous, security-minded, and open to technical challenge.

Evening Star AI connects technical research with experimental systems and published analysis. Questions, corrections, and reproducible results help improve the work.

View LinkedIn Discuss the Research