Profile
I help teams decide whether technically ambitious ideas can work, identify the uncertainty that matters and turn it into an architecture, experiment, prototype or working system. My background spans software engineering supporting NASA Ames via Amentum, computational neuroscience, applied AI, original algorithm development and scientific systems. I hold an M.S. in Mathematics and a B.S. in Computer Science; my first-authored 2024 paper developed a neural-network and Shapley-value method for constructing shorter psychometric measures. I work directly across research and implementation, making assumptions testable and leaving teams with evidence they can use.
Consulting Focus
I am most useful when the goal matters but the technical path is unclear: feasibility and architecture investigations, applied AI and LLM systems, bounded prototypes, algorithmic or mathematical R&D, scientific software, evaluation and benchmarking, and independent technical review. I also provide fractional R&D direction when a team needs sustained technical guidance across experiments, implementation choices and changing evidence. Where specialist contributions are needed, I can scope, direct and integrate them.
Selected Work
Nexus — AI operations and exact action authorization
Developed Nexus from a local prototype into a human-governed operational platform integrating Google Workspace, web research, social scheduling, cloud infrastructure, Neo4j/Aura and affiliate tooling.
- Designed the boundary between model reasoning and operational execution: a typed operation registry and plan compiler validate proposed actions, and human approval binds the exact reviewed content to a scoped dispatch permit.
- Executed real Gmail delivery and recipient receipt, read-only inbox ingestion, reply matching and response drafting, alongside live web-research enrichment and operational Neo4j/Aura schema work.
- Live-validated a controlled Buffer lifecycle—schedule, read-back, idempotent reuse and deletion before publication—with approval gates, pacing, suppression, audit trails and emergency-stop controls around external actions.
AWS delivery and queue-event infrastructure was exercised but is not the current outbound route; bounded Tapfiliate reads and test writes were validated while general affiliate operations remain gated.
Correspondence Matrices — Boolean expression evaluation
Turned an original mathematical representation of Boolean functions into public, inspectable software with reproducible comparisons and independent correctness checks.
- Designed and built Python compilers around a shared symbolic intermediate representation, with eager, lazy and hybrid evaluation paths. Structural normalization, hashing and caching support reuse of compiled expressions.
- Built a reproducible evaluation harness that separates compilation from execution, checks results against independent truth-table evaluation and publishes per-trial data and consolidated reports. This makes both the idea and the limits of the measurements open to inspection.
Public implementation on GitHub · Published results and audit · Working paper: Correspondence Matrices; Algorithms for Propositional Logic
Paper Workbench — evidence-controlled research and publishing
Conceived, architected and built a public, local-first research environment through an AI-assisted implementation workflow, retaining responsibility for review, testing and capability boundaries.
- Connected sources, excerpts, claims, dialogue, manuscript sections and exports in a provenance graph with explicit support states, claim-to-evidence links, human-gated AI actions and audit-in-transaction.
- Implemented literature discovery, manuscript and artifact workflows, deterministic integrity audits and provenance-bearing export. The published record includes 101 offline tests, a seven-code audit evaluation, a bounded live-provider behavior suite and explicit limitations.
Public repository · Capability matrix · Evaluation reports
Fractilate — mathematical art and encoded images
2019 – present · fractilate.com
A public mathematical-visualization platform that turns equations, fractals, attractors and landscapes into an interactive creative medium. The work connects mathematical modelling, rendering systems and product development across several runtimes.
- Developed web, Python desktop and browser-studio implementations of a custom rendering core, using parity tests to protect shared visual intent across implementations.
- Built composition scoring and preference-learning workflows to organize a large generative space, while keeping final selection with the user. Added recoverable message encoding using a variant of the Correspondence Matrix representation, with a public decrypt tool and compatible Node/.NET encoding.
Presidency or Prison — game simulation and development
2024 – present · presidencyorprison.com
A strategy-card-game project developed as an integrated rules, simulation, computational-design and media-production system. The work makes changing relationships between cards and visual forms testable rather than relying only on isolated values or intuition.
- Built a Python simulation engine using NumPy, pandas and NetworkX, with structured configuration and an operator CLI, and developed a digital-game implementation in Dart.
- Developed CCSGL, a constrained geometric symbol language, by enumerating candidate forms, measuring separation, optimizing a provisional codebook and adversarially checking initially plausible design assumptions. Human-recognition validation remains incomplete.
- Built traceable AI-assisted media workflows in which automated analysis proposes structure, scenes or cuts and human editorial feedback determines what is accepted. Examples: What POP Is and How POP Achieves.
Across Nexus, Fractilate and Presidency or Prison, research, media and workflow tools are infrastructure built to support the products.
Personal Projects
B-Theory — mathematical research on self-awareness
A long-running exploratory research programme investigating mathematical representations of state, distinguishability and self-reference. The work connects Boolean algebra, topology, homology and continuous change, asking how a system might represent and compare its own state. It is an open research programme, not an established model of consciousness.
- Developed four connected research directions—Algebraic Logic, Logical Topology, Logical Homology and Evolutionary Logic—to move the question between symbolic, spatial and dynamic representations.
- Developed Correspondence Matrices within this programme, then separated the idea into its own computational project with a public implementation, working paper and reproducible evaluation tooling.
B-Theory research site · About and project notes
Experience
Independent R&D — product development and contractor direction
Built the independent products and research programmes above while directing specialist contributions where appropriate. Hired and technically directed dozens of freelance specialists across more than 100 engagements, including extensions and recurring collaborations lasting multiple years. Scoped and reviewed software, design, research and media work while retaining responsibility for architecture, technical decisions and integration.
NASA Ames (Via Amentum) — Software Engineering Lead
2022-12-11 through 2025-09-15
Led software engineering supporting Thermophysics Research across local-AI information systems, numerical software, instrumentation and engineering integration.
- Developed local-LLM, NLP, retrieval and knowledge-management workflows for technical document analysis and information extraction.
- Built numerical software and integrated communications infrastructure, PLCs, instrumentation and high-speed data acquisition across software and hardware boundaries.
- Coordinated integration, documentation and technical reviews supporting Critical Design Review and System Integration Review readiness.
Selected project accounts: PROFINET data-path engineering · Local multi-agent AI for engineering analysis · Quasi-1D thermophysics modeling
Palo Alto University — Computational Neuroscience Research Associate
July 2018 – July 2022
Developed machine-learning and neuroimaging methods for problems in measurement and brain-network analysis.
- Designed and built neuroimaging processing and analysis pipelines in MATLAB and FSL. Co-authored the resting-state network study cited below.
- Developed a neural-network and Shapley-value method for selecting questionnaire items for shorter psychometric measures, evaluating retained configurations against explicit score targets.
- Investigated how item selection changes with the intended measurement target. This work informs my approach to AI evaluation: define the decision target first, then check what the reported result actually measures.
Publications arising from this research
- Droncheff, B., Liu, K., & Warren, S. L. (2024). Item Retention as a Feature Selection Task: Developing Abbreviated Measures Using Shapley Values. Journal of Psychopathology and Behavioral Assessment, 46, 220–234. First author.
- Warren, S. L., Geller, W. N., Droncheff, B., & Liu, K. (2025). Individual differences in anxious apprehension and anxious arousal alter resting-state network structure and connectivity. Journal of Affective Disorders Reports, 20, 100895. Co-author.
Early Research
NASA Jet Propulsion Laboratory — 2003–2006; 2008–2009. Built LabVIEW drivers and I/O for physics instrumentation and laboratory automation; developed adaptive algorithms and scientific software in Java, C and MATLAB.
TU Berlin — Summer 2006. Developed C++ libraries supporting molecular beam epitaxy research.
California Institute of Technology — Summer 2004. Built computer-based experiments studying human rationality and decision-making.
Additional Technical Experience
Technical Tutoring — 2004–2011. Tutored 100+ people in mathematics, physics, computer programming and web design, concurrent with university and research work.
Education
M.S. Mathematics — ALGANT Erasmus Mundus Program, University of Padova and Chennai Mathematical Institute, 2015. Dual-degree programme in algebra, geometry and number theory.
B.S. Computer Science, Minor in Mathematics — California State Polytechnic University, Pomona, 2011.
Selected Technical Skills
Languages: Python, TypeScript/JavaScript, C++, C#, Java, Dart, MATLAB, SQL, LabVIEW.
AI and systems: LLM integration, retrieval, governed agent workflows, development-control architecture, authorization and auditability, evaluation and benchmarking; FastAPI, Next.js/React, PostgreSQL, Neo4j, Docker.
Scientific computing and verification: NumPy, SciPy, pandas, NetworkX, FSL; pytest, Vitest, Playwright.
Discuss a Project
Research identity: ORCID 0000-0001-9332-8934.
For a scoped investigation, prototype or ongoing R&D engagement, use the contact form with the problem, the constraints and the decision you need to make.