"Design top-down. Verify bottom-up. Release with evidence."
Creativity and technical rigor are complementary phases of the same engineering process.
I build deterministic, governed computational systems where AI, physical systems, and high-liability decision-making intersect. My work asks how a system can remain understandable and safe under real constraints: make state explicit, define boundaries, constrain unsafe transitions, preserve fallback paths, and leave an audit trail.
This profile is the front door to an independent research program spanning software, hardware, mathematics, and scientific tooling. Across those domains, the goal is consistent: turn clarity, determinism, trust, reproducibility, auditability, and safety into properties of the architecture—not assumptions about its operators.
The repositories apply a common systems philosophy at different layers:
- Governance and context — trust membranes, attributable context transfer, policy constraints, and bounded behavior for agentic systems.
- Runtime and verification — deterministic pipelines, distributed and verifiable computation, reproducible evaluation, and explicit failure handling.
- Physical and embedded systems — computer vision, smart glasses, EV and battery systems, biosignals, and other interfaces where computational decisions meet the physical world.
- Mathematical and scientific foundations — nonlinear dynamics, phase memory, invariant-driven modeling, simulation, and falsifiable reference implementations.
- Interoperability — practical bridges between legacy and modern compute without hiding system state or weakening governance boundaries.
These are not disconnected projects. They explore the same question from different directions: how can complex systems act usefully while remaining inspectable, constrained, recoverable, and worthy of trust?
flowchart LR
A[Governance] --> B[Context]
B --> C[Runtime]
C --> D[Verification]
D --> E[Physical Systems]
D --> A
Governance defines permitted action; context supplies attributable state; runtimes execute within those boundaries; verification closes the loop; and physical systems provide the hardest test of whether the architecture behaves as intended.
| Repository | What it explores |
|---|---|
| AILEE Trust Layer | An adaptive trust and governance middleware architecture that evaluates AI-generated decisions before they become system actions. Integrates deterministic trust scoring, confidence thresholds, contextual mediation (GRACE), distributed consensus, auditable decision pathways, and stability-preserving fallback mechanisms across multiple computational and industrial domains. |
| Ambient AI + VCP System | A distributed, verifiable computing architecture coordinating heterogeneous devices, AI workloads, and decentralized processing resources through a unified orchestration framework. Combines node-health intelligence, trust-based task scheduling, cryptographic verification with zero-knowledge proofs, and resilient offline-first operating capabilities. |
| AeroCam | A multidisciplinary C++ engineering framework exploring real-time motion intelligence, embedded control systems, inertial sensor fusion, and computational camera stabilization. Integrates ARM Cortex-M firmware, Extended Kalman Filtering, OpenFX video-processing tools, and experimental motion-aware computer vision applications for aerospace, filmmaking, and augmented reality. |
| ISLA | An experimental digital audio signal-processing platform investigating phase-aware note smoothing, harmonic blending, and perceptually connected musical transitions. Combines offline WAV processing, real-time macOS system-audio monitoring, and interactive terminal and desktop interfaces with Velvet, Legato, Melt, and transparent Bypass listening modes. |
| U.F.O. | A governed, compute-aware AI architecture modeling adaptive behavior and computational constraints through a deformable radial membrane and dynamically routed V-Channels. Integrates bounded reasoning, geometric stability control, behavioral governance, trust memory, adaptive recurrence, and auditable simulation to investigate computationally responsible AI personalization. |
| Context Broker | A vendor-neutral context-interoperability framework supporting controlled information exchange between AI systems, software applications, and connected devices. Uses explicitly authorized, time-bounded, attributable context packets with deterministic expiration, provenance tracking, and ALCOA-oriented auditability to reduce uncontrolled persistence and cross-system ambiguity. |
| DS-EV | A research-stage electric-vehicle battery intelligence and energy-governance architecture investigating dual-state battery modeling, degradation estimation, regenerative-energy management, and battery-aware power control. Combines C++/Python simulation, safety-boundary enforcement, energy-continuity accounting, and auditable BMS middleware, with physical vehicle validation remaining outside its established capabilities. |
Where to start: begin with AILEE Trust Layer for the governance model, then follow Context Broker and Ambient AI + VCP System through context, execution, and verification. AeroCam, Trust Beacon, and DS-EV show those principles applied at physical-system boundaries; Phase-Memory Operator provides a mathematical lens on state change and memory.
I favor systems that are:
- Interpretable — important state and decision boundaries are visible.
- Deterministic where it matters — identical inputs and conditions support reproducible evaluation.
- Auditable — actions, transitions, and provenance leave inspectable records.
- Falsifiable — research claims can be tested against simulations, benchmarks, or explicit invariants.
- Fail-safe — unsafe transitions are constrained and fallback paths remain available.
- Composable — governance, context, runtime, and verification can evolve without collapsing their boundaries.
The work is research-stage unless a repository states otherwise. I use terms such as reference implementation, simulation, and validated benchmark deliberately so that exploratory results are not mistaken for production claims.
- Quality, Manufacturing & Validation — Merck, Accupac, and Microsize.
- Systems & Research Writing — Orbital Robotics, including ITAR-compliant SBIR technical work.
- Independent Research — agentic-AI safety, biosignal modeling, deterministic systems, and reproducible scientific infrastructure.
- Pennsylvania State University — B.S. in Business, with a concentration in Marketing and Management.
- University of Pennsylvania — pursuing an M.S. in Computer Science beginning January 2027, with a focus on AI.
- Financial supporter of the World Food Programme.
- American Red Cross platelet donor, with 40 units donated.
- Supporter of the Chopan Foundation, serving children and families facing hardship in the Middle East.
- Research discussion and collaboration: open an issue in the relevant repository.
- Open source: research repositories are MIT-licensed where noted.
- Email: dfeen87@gmail.com