ActiveDealership operations platform
Wheelbase
Auction intelligence, inventory, and recon — one Go backend, shipped to web, desktop, and mobile.
Open live sitewheelbase.io
Available · Boston, MA·
Matt here — I write software meant to hold up under real use, not just look good in a demo.
Building Wheelbase, a dealership operations platform. Creator of Grammario, a grammar analyzer for six languages.
Available for new opportunities — reach out any time.
GitHub has the code, the journal has the write-ups, and my inbox is always open — say hello.
Work
Things that had to ship.
Two live products. Six completed builds. GitHub is the profile, not a set of invented repo names.
ActiveGrammar you can see
Grammario
Click a sentence. See the structure. Universal Dependencies first, AI explanation second.
Read the case studygrammario.ai/analyze
Click a word
Past participle
Agrees in gender with the clitic. The structural head of the clause.
Completed
CEFR graded readers
Levla
Graded-reader app for Russian and Japanese, A1–B2. An LLM drafts each passage under CEFR-level grammar constraints, a morphological analyzer validates and rewrites it before it ships, and tapping any word surfaces lemma, grammar, gloss, and kanji readings. Feedback on each text adjusts placement and recommends what to read next.
Next.js · FastAPI · SQLAlchemy · Sudachi · pymorphy3 · OpenRouter
Automated GitHub changelogs
Purser
A read-only GitHub App that turns each merged pull request into a reviewed changelog draft: dependency bumps and CI-only diffs are filtered out before any model call, then Claude writes a customer-facing note and a technical note per PR. Approved entries push to Slack and an in-app launcher widget, with Stripe-metered plans and Upstash QStash jobs so it runs serverless on Vercel.
Next.js · Supabase · GitHub Apps · Upstash QStash · Anthropic API · Stripe · TypeScript
3D globe and dashboard
Global Terrorism Visualization
Interactive WebGL globe over 177,000+ records from the Global Terrorism Database, plus a seven-view analytics dashboard covering trends, regions, attack types, targets, weapons, and hotspots.
React · Globe.gl · FastAPI · Pandas · Vite
Behavioral ML pipeline
Teen Phone Addiction Prediction
Random forest regressor predicting adolescent phone-addiction risk from behavioral survey data, tracked with MLflow and served through a FastAPI endpoint that always loads the latest registered model, with a Streamlit dashboard for exploration and live prediction.
Python · scikit-learn · MLflow · FastAPI · Streamlit
Lesion classification
SkinGuard
Binary skin-lesion classifier (benign vs. malignant) trained on HAM10000, with four training pipelines across TensorFlow and PyTorch and export to Apple Core ML for on-device iOS/macOS inference.
TensorFlow · PyTorch · coremltools · MLflow · scikit-learn
Linux process monitor
Procmon
Terminal process monitor in C++ with ncurses. Real-time procfs reads for CPU, memory, and threads, plus sorting, filtering, and signal dispatch.
C++ · ncurses · Linux · procfs
PythonGoTypeScriptC++ReactNext.jsFastAPINode.jsTensorFlowPyTorchscikit-learnspaCySupabasePostgreSQLDockerLinux
Experience
Where the work happened.
2024 — Present
Full-Stack Engineer · Wheelbase
Multi-tenant dealership operations platform built from scratch as a Turborepo monorepo: web app, Electron desktop, and an Expo mobile field app sharing one Go backend. Re-engineered the backend from Python to Go for sub-second response times, and built an AI operations assistant with tenant-scoped natural-language SQL.
Mar – Sep 2025
AI / Data Science Intern · MA Executive Office of Administration & Finance
Applied machine learning and NLP to state government data pipelines, and delivered briefings on AI procurement risk to senior staff.
Education
Boston University
M.S. Applied Data Analytics
2026 — 2028 (expected)
University of Massachusetts Lowell
B.S. Computer Science — Data Science Concentration
2020 — 2024
Certifications
- Neural Networks and Deep Learning — DeepLearning.AI
- NLP Specialization — DeepLearning.AI
- Google Go Programming Specialization — Google / Coursera
About
Engineer, not a generic AI founder.
Software engineer in Boston, MA. Building Wheelbase, a dealership operations platform. Creator of Grammario, a grammar analyzer for six languages. Applied machine learning to state government data pipelines at the MA Executive Office of Administration & Finance.
I like hard problems and am drawn to domains where getting it wrong has real consequences.
I am interested in systems that have to work under pressure — embedded, constrained, or otherwise consequential. On the language side, Grammario came from a frustration with how grammar is usually taught: I wanted to see the structure, not memorize rules. That turned into a real NLP pipeline covering morphological analysis, dependency parsing, and the structural differences between agglutinative and fusional languages. The seven languages I study inform the engineering as much as the other way around.
Languages
Seven languages, seven different grammars.
Grammario exists because I wanted to see structure, not memorize it. These are the languages that shaped how it works — read the case study.
English
Native
Russian
Native
Italian
Proficient
Turkish
Proficient
German
Basic
Hebrew
Basic
Japanese
Basic
Journal
Public writing.
Build notes. Mostly Grammario, plus the Go work that came out of a slow Wheelbase pipeline.
Grammario: Where Things Standv1.0 features, teacher suite, Japanese, Learn section. Honest about building solo.Why I'm Changing Everything: The New Grammario MethodStructural-First Analysis. Analyst, Strategist, Tutor. Language-family strategies.Grammario DownBackend maintenance. Analysis API temporarily down.My Thoughts on GoPointer and method-receiver syntax versus C/C++. Automatic struct dereference.
See all 12 entries →Contact
Want to get in touch?
Available in Boston, MA. Reach out if you're hiring, or just want to talk shop.