Available for new engagements

Designing the architecture
behind modern software.

Full-stack development and AI agent systems for teams who need more than a template — from data model to deployed product, designed to hold up as it grows.

View the work ↓
Currently booking projects starting Q4 2026
React React Node.js Node.js Python Python FastAPI FastAPI LangChain LangChain MongoDB MongoDB PostgreSQL PostgreSQL Docker Docker Playwright Playwright

Services

Where I plug in

Engagements usually fall into one of these — often more than one at once, since the interesting problems sit at the seams.

01

Full-stack web apps

End-to-end product builds on the MERN stack, from data model and auth through a shipped, maintainable UI.

ReactNodeMongoDB

02

AI agent systems

Orchestrated, multi-agent backends — scoped, testable, and built to fail gracefully instead of silently.

LangGraphFastMCPPython

03

Automation & tooling

Browser automation, internal CLIs, and scripts that remove the parts of your workflow that shouldn't need a human.

PlaywrightCLIScripting

04

Architecture & review

A second set of senior eyes on system design, API shape, and technical debt before it becomes expensive.

API designAuditsScaling

Selected work

A few things I've shipped

A cross-section of recent projects — client names withheld, problems and stacks intact.

saas · automation

CASE STUDY

Browser automation platform

A hybrid deterministic/agentic system that runs client-side, cutting server compute costs to near zero.

PythonPlaywrightFastAPI
agents · orchestration

CASE STUDY

Multi-agent orchestration framework

A three-tier agent hierarchy — orchestrator, sub-agent, task agent — with a decorator-first API.

LangGraphLangChainFastMCP
web app · mern

CASE STUDY

Client operations dashboard

A MERN-stack internal tool replacing three spreadsheets and a shared inbox with one source of truth.

ReactExpressMongoDB
cli · tooling

CASE STUDY

Media extraction CLI

A fast, scriptable command-line tool for batch media extraction, built with a clean progress-first UX.

PythonRichyt-dlp
systems · low-level

CASE STUDY

Terminal graphics engine

A from-scratch turtle graphics library with Bresenham rasterization, for teaching low-level rendering.

CPythonRasterization
ml · research

CASE STUDY

Transformer, implemented from scratch

A ground-up PyTorch implementation of the Transformer architecture, built to study — not just use — attention.

PyTorchPython

Process

How an engagement runs

The same four stages on every project, so you always know what's next and why.

01 — Discovery

Understand the problem

A working session on goals, constraints, and what "done" actually means before any code exists.

02 — Architecture

Design the system

Data model, API shape, and the boundaries between services — decided on paper, not in production.

03 — Build

Ship in slices

Working software in short cycles, with a demo at the end of each one instead of a single big reveal.

04 — Handoff

Leave it maintainable

Documentation, tests, and a walkthrough — the system stays understandable after I'm off the clock.

About

Built for the parts other developers avoid

I'm a software engineer specializing in custom software, AI integration, and business automation. I work with businesses that need solutions beyond off-the-shelf software, designing applications and workflows tailored to the way they operate. The result is technology that reduces repetitive work, improves efficiency, and grows alongside the business.

My engineering foundation comes from systems programming, algorithm design, and building technology from first principles. That experience shapes how I approach modern AI and software development: designing scalable architectures, optimizing intelligent systems, and creating automation solutions that deliver real business impact. Clients work with me when their challenges require more than another feature—they need technology built to evolve.

7+years building software
20+shipped projects
3core stacks: MERN, Python, agents

Testimonials

What clients say

A few words from people I've built with recently. Names shown are placeholders — swap in real quotes and client photos as they come in.

Handed over a vague problem and got back a system we actually understood — not just code that worked, but a reason for every decision in it.
PA

Priya Anand

Product Lead, Fenwick Analytics

The agent framework held up under load we hadn't even planned for yet — the kind of margin you only get from someone who designs for scale, not just the demo.
ME

Marcus Ellery

CTO, Northline Logistics

We stopped losing hours a day to manual data entry within the first two weeks. It just works, and nobody on the team has had to think about it since.
DW

Dana Whitfield

Operations Director, Larkspur Studio

Contact

Tell me about the project

The more specific, the better — stack, timeline, and what "success" looks like all help me reply with something useful.

RESPONSE Within 1–2 business days
STATUS Booking for Q4 2026
ELSEWHERE GitHub  ·  LinkedIn  ·  Upwork

Sent straight to my inbox — no auto-reply spam, just a real answer.

Message received — I'll reply within 1–2 business days.