Research & Publications

Two published books, applied research on neural simulation at scale, and open-source frameworks maintained in public — plus the reading list and toolbox behind the work.

Author

Published Books

Two books documenting the methodology behind the supply-chain and world-economy simulation work — how multi-layer perceptrons can model systems as large as an international supply chain or the global economy itself.

Simulating the Entire World Economy with Multilayer Perceptrons

ISBN: 979-828080999-4

The world-economy simulation work — how a hierarchical multi-layer perceptron architecture can model the interacting economies and sectors of the entire globe.

View on Amazon

Creating a Simulation of an Entire International Supply Chain Using Multi-Layer Perceptrons

ISBN: 979-828041476-1

The methodology behind the 25-module supply chain simulator — modeling an international supply chain end to end with deep neural networks.

View on Amazon
Papers & Research

Applied Research

Research on what multi-layer perceptrons can be pushed to do at scale — from computing itself to macroeconomic forecasting and large language models built from scratch.

Realizing a Fully Functional CPU Using Multi-Layer Perceptrons

A paper on neuromorphic computing: a fully functional CPU realized with 25M+ perceptrons across 3k+ layers — computation itself, rebuilt from neurons.

Read the Paper (PDF)

H-MLP World Economy Model

Published research: a 23M-parameter hierarchical MLP simulating 155 economies and 45 sectors, beating VAR-X, DSGE, and Transformer benchmarks by 30% over 12-quarter forecasts.

Impersona

A 2B+ parameter character-level Transformer LLM with custom multi-head attention, built from scratch and used to produce long-form content for 7 client projects.

Open Source

Frameworks I Build in the Open

Five projects I maintain publicly — from securing AI agent systems and benchmarking quantum finance algorithms to a Rust plotting library and a Hinglish programming language, part of 36 public repositories and 16 published packages across PyPI, npm, and crates.io.

AASTF — Agentic AI Security Testing Framework

GitHub stars for AASTFAASTF license badge

A Python framework for security-testing AI agent systems — the model plus its tools, memory, and planning loops, not just prompts. Ships with 130+ built-in attack scenarios mapped to the OWASP Top 10 for Agentic Applications, VULNERABLE / REFUSAL_ECHO / SAFE verdicts, sandboxed tool backends with no side effects, a LangGraph adapter plus experimental CrewAI, OpenAI Agents, and PydanticAI adapters, 25 MCP security scenarios, SARIF output for CI/CD, and EU AI Act readiness scoring. MIT licensed.

View on GitHub Project details

qufin — Quantum-Enhanced Quantitative Finance

GitHub stars for qufinqufin license badge

An open-source framework where every quantum algorithm ships alongside the best classical solver for head-to-head benchmarking: 159 modules, 14 subpackages, and 11 backends including Qiskit, PennyLane, Cirq, Braket, CUDA-Q, D-Wave, IonQ, and Quantinuum. Covers portfolio optimization (QAOA, VQE, annealing vs Mean-Variance, Black-Litterman, HRP), option pricing with six QAE variants vs Black-Scholes and Monte Carlo, quantum VaR via HHL, and eight error-mitigation strategies. One command away: pip install qufin.

View on GitHub Project details

aiguard — Security Harness for CLI Coding Agents

GitHub stars for aiguardaiguard license badge

A portable security harness for CLI coding agents, written in Rust as an 11-crate workspace — it sits between the agent and your local system, intercepting tool calls with layered checks: prompt-injection detection (80+ rules with zero-width Unicode and base64 payload checks), secret redaction (52 gitleaks-compatible patterns with entropy gating), MCP tool pinning with SHA-256 pins and rug-pull alerts, shell/path deny rules, tamper-evident audit logging, and a session-replay TUI. Supports 8 agents including Claude Code, Codex CLI, Gemini CLI, Aider, and Goose — dual-licensed Apache-2.0/MIT.

View on GitHub Project details

plotkit — Pure-Rust Plotting Library

GitHub stars for plotkitplotkit license badge

A pure-Rust plotting library — publication-quality plots in three lines of Rust, with a matplotlib-familiar API (pyplot facade plus a Figure/Axes model), 16 chart types, and PNG/SVG/PDF output. Ships with 7 themes, 14 perceptual colormaps, first-class Polars and ndarray integration, a WASM/canvas backend, and an embedded font for deterministic rendering — no GPU or system dependencies, dual-licensed MIT/Apache-2.0.

View on GitHub Project details

adarshLang — A Hinglish Programming Language

GitHub stars for adarshLang

A Hinglish-flavoured programming language built for fun demos and workshops, with keywords like badlo (variables), dikhao (print), and agar/warna (if/else). The full pipeline — lexer, parser, semantic analyzer, and interpreter — is written in Python, with a REPL and a live Flask web playground.

View on GitHub Try the Playground
Reading

What I Recommend Reading

Books I keep coming back to — and recommend to clients more often than any tool or framework.

The Pragmatic Programmer

Hunt & Thomas — The mindset book. Most of what clients pay consultants for is in here: tracer bullets, good-enough software, owning your craft.

Designing Data-Intensive Applications

Martin Kleppmann — The clearest thinking in print on databases, replication, and the trade-offs behind every architecture decision I make.

The Web Application Hacker's Handbook

Stuttard & Pinto — To defend an application you have to understand how it gets attacked. Still the foundation of every security review I run.

The Phoenix Project

Kim, Behr & Spafford — A novel about IT that every non-technical founder should read. Explains why work-in-progress is the silent killer of delivery.

Co-Intelligence

Ethan Mollick — The most level-headed take on working with AI — useful for deciding what to automate and what to keep human.

The Mom Test

Rob Fitzpatrick — How to ask questions that get honest answers. I use it in every discovery call before writing a line of code.

Toolbox

What I Work With

The tools I reach for every day.

Frontend

React, Next.js, TypeScript, Tailwind — fast interfaces that hold up under real users.

Backend

Node.js, Python (FastAPI/Django), PostgreSQL, Redis — boring, proven, maintainable.

AI / LLM

Claude & OpenAI APIs, RAG pipelines, embeddings, and custom Transformer models — with evaluation harnesses for production AI.

Automation

Workflow orchestration, queues, webhooks, and integrations that glue your systems together.

Cloud & DevOps

AWS, Docker, CI/CD, infrastructure-as-code — deploys you can trust on a Friday.

Security

OWASP tooling, agentic AI security testing, dependency scanning, threat modeling, secure code review.

Data & Quant

ETL pipelines, analytics dashboards, and quantitative finance models — valuation engines, derivatives analytics, and quantum-enhanced optimization.

Delivery

Weekly demos, written decision logs, and documentation your next hire can actually onboard from.

Want to put this research to work?

Tell me what you're building — AI systems, quantitative models, or something in between. First consultation is free, and I reply within 24 hours.