Boston Rudi — Full-Stack Founder & AI Engineer

Portfolio · Open Call — Mercury · Full-stack Founder + AI Engineer

Shipping AI features
in production,
not in theory.

I'm Boston Rudi — I built ContractorOS, a production SaaS platform with live AI features: document OCR, PDF scorecard parsing, and a proxied Anthropic API serving real users. I direct the product; I ship the code. Phase 2 adds four autonomous agents. No prior CS background. Just a problem, a platform, and a lot of iteration.

contractoroshub.com — live

ContractorOS LLC

Anthropic API in production

UAT in progress

React · Vite · Supabase · Clerk · Stripe

The product

The operating system
for contractors

ContractorOS serves five contractor segments — OTR, FedEx Ground ISP, Amazon DSP, last-mile delivery, and USPS HCR — each with its own screen set, compliance rules, and pay structure. Built segment-awareness in from day one because I understood the operators.

Contractor segments

30+

Screens in production

50+

Supabase data keys, org-scoped

Stripe-gated pricing tiers

AI features — all proxied, none exposed

Document OCR, Amazon Scorecard PDF parsing (DART/DCR/Mentor), load analysis. All calls routed through /api/claude.js — the browser never touches the API key. Learned this the hard way: caught and fixed a real key exposure, rebuilt the pattern permanently.

Anthropic APIServerless proxyPrompt engineering

Multi-tenant auth, membership-gated

Every user belongs to a Clerk Organization. All Supabase data scoped to org ID. Membership-required mode enforced at app level. Owner bypass for admin access. Role system tied to org membership.

ClerkSupabase RLSMulti-tenancy

Stripe payments, end to end

Three products in sandbox. Checkout session via serverless function (CommonJS — learned after a 405 error). Webhook writes tier back to Supabase via service role key. Locked nav items open upgrade modal. Success redirect updates tier on return.

StripeWebhooksTier gating

Full lender report suite

EBITDA waterfall, 12-month cash flow table, DSCR, auto-generated balance sheet, debt schedule, AR/AP aging, assets & liabilities, document checklist with IRS/SBA/FMCSA links. Enterprise-only.

EBITDADSCRFinancial modeling

Live FMCSA carrier lookup

Real-time safety snapshot vs national OOS averages. Renewal calendar auto-populated from lookup results. Segment-specific compliance checklists pre-loaded per contractor type on first login.

FMCSA APIComplianceDOT

Business Health Score

0–100 composite across four pillars: Compliance, Financial, Drivers, Fleet. 8-week sparkline trend. Lender Ready indicator (0–100) pulled from the same data. At-a-glance operator view — no natural language, just numbers that matter.

Dashboard logicTrend analysis

Segments & pricing

Five segments,
one platform

Each segment gets its own screen set. The platform detects segment on sign-up and surfaces only what's relevant — no noise for a USPS driver, no missing tools for an Amazon DSP owner.

Segment	Segment-specific screens
OTR / Owner-Operator	Dead Miles · Load Analyzer · Load History · Fuel Card Reconciliation
FedEx Ground ISP	Stop Profit · Settlement · Driver Schedule · Scorecard · Coaching Log · Appearance Tracker
Amazon DSP	Stop Profit · Settlement · Driver Schedule · Scorecard · DNR Cases · Van Inspection · Callout Tracker
Last Mile / Lowe's	Stop Profit · Scorecard · Claims Manager · White Glove Checklist
USPS HCR	Scorecard · Trip Sheets · HCR Bid Tracker · Sub Driver Approval Log

Solo

$39/mo

Compliance · Fleet · Finance · Docs · FMCSA

Fleet

$89/mo

+ Drivers · Payroll · Dispatch · Settlements · Scorecard

Enterprise

$179/mo

+ Lender Suite · Analytics · Autopilot agents

Phase 2 — Autopilot

Four agents.
Zero input required.

ContractorOS Autopilot is a Python-based agentic backend that monitors contractor data and acts autonomously. Enterprise add-on and AI engineering portfolio project — same build, two purposes.

Compliance Monitor

Polls FMCSA weekly per DOT number, compares to stored dates in Supabase, sends SMS and email if anything changed or approaching expiry. Never miss a renewal.

FMCSA API · Supabase · Resend · Twilio

FMCSA Sync Agent

Pulls live inspection and safety data weekly, auto-updates ContractorOS compliance records, flags new violations before they become OOS events.

FMCSA · Supabase Python client

Settlement Watchdog

When a new deposit is logged, compares to expected settlement. Texts contractor immediately if variance exceeds threshold — like an accountant watching every Friday deposit.

RAG · pgvector · Twilio

Weekly Briefing Agent

Every Monday at 6am, generates a plain-English business summary and texts it via Twilio. Revenue, compliance status, driver issues, what needs attention this week.

LangChain · Claude API · Twilio

Python + APIs

FMCSA calls · Supabase reads · Resend emails

First agents live

A1 + A2 deployed to Railway

Orchestration

LangChain · multi-agent memory · A4

RAG pipeline

pgvector in Supabase · A3

Production

Evals · observability · reliability

Apply

AI Engineer roles · $130K–$250K

Stack

Everything
in production

React + Vite

Frontend

Supabase

Database + storage

Clerk

Multi-tenant auth

Stripe

Payments + webhooks

Anthropic API

AI features, server-proxied

Vercel

Hosting + serverless functions

FMCSA API

Live compliance data

Resend + Twilio

Email + SMS (Autopilot)

LangChain

Agent orchestration (planned)

pgvector

RAG pipeline (planned)

Why Mercury

How I map
to what you're building

On "find the problem before anyone assigns it"

I didn't get handed a product brief. I saw a market of contractors running their entire business in spreadsheets and built the tool they needed. The problem was obvious to someone close to the operators. That instinct — staying close enough to spot what matters — is how ContractorOS was designed from day one.

On "build and automate end to end using our AI stack"

I have AI features in production right now — document OCR, PDF scorecard parsing, load analysis. I've handled a real security incident (key exposure, unauthorized charges, rebuilt the pattern). Phase 2 is four autonomous agents watching contractor businesses 24/7. I know how to move an AI idea from concept to deployed endpoint.

On "ship fast, measure honestly, kill what didn't move"

I removed Ask DOT AI and natural language search after recognizing operators need at-a-glance dashboards, not chat. That call saved API costs and made the product tighter. I don't polish what doesn't move the needle. I've made that call on real features, not hypothetical ones.

On "dangerous across the stack"

I started with zero coding background. I now direct architecture, catch security vulnerabilities, debug production issues, and ship weekly. I learned by building something real that real people will pay for. ContractorOS is the portfolio and the proof.