Jonathan Cruz

Queens, NY

Jonathan David Cruz

Nearly 3 years building backend systems and automating manual workflows, including an internal Claude-based AI agent that handles support tickets. Now focused on software, AI, and automation engineering roles.

About

I spent nearly three years as a Software Administrator (functionally a software engineer) at Friedman Vartolo LLP, a law firm, building backend systems with Python, SQL, FastAPI, and Azure. I co-built an internal Claude-based AI agent that matches new tickets against 2,000+ past ones and carries out approved tasks through Playwright, and built an automation dashboard that saved the firm 479+ hours of manual entry. I also trained an Azure Document Intelligence model to pull data out of vendor invoices. Outside of work I build things end to end, like Riffstub, a concert tracking app I designed and shipped solo, and its native Android app, which I'm building now.

I graduated from SUNY College of Old Westbury with a B.S. in Computer and Information Science in 2022.

Experience

Software Engineer, AI & Automation (Software Administrator)

Dec 2023 – Oct 2026

Friedman Vartolo LLP · Garden City, NY

  • –Co-built an internal AI agent on Claude: extracted 2,000+ historical tickets and how each was resolved into JSON so Claude could match new tickets to past solutions, and wrote the prompts that turn those matches into step-by-step instructions. The agent confirmed its instructions were correct by querying the firm's database to match them against the expected output. Built Playwright automations to carry out approved instructions, and validated outputs through human review; it reliably resolved about 70% of routine tickets and gave staff a head start on the rest.
  • –Built an automation dashboard that let the team classify incoming case updates from a third-party servicing platform and automatically apply handling rules, eliminating manual entry for the boarding team. Automated 1,495 workflows across 71,000+ case updates, saving 479+ hours.
  • –Built a Streamlit monitoring dashboard, backed by FastAPI and deployed to Azure via Docker, to track production automations. Configured automated failure alerts that emailed me and the affected team immediately so they could act while I deployed a fix.
  • –Trained a custom Azure Document Intelligence model to extract data from incoming vendor invoices into structured JSON by labeling sample documents, then reconciled the extracted fields against internal case and payment records to confirm payments and flag discrepancies.
  • –Streamlined document prep by integrating SQL queries with dynamic templates to auto-generate case-specific court filings and client documents (motions, affidavits) as cases progressed through defined steps.
  • –Partnered with non-technical staff across departments to train them on new automation and AI tools, writing clear internal documentation for every tool shipped.

Projects

Next.js · TypeScript · React · PostgreSQL · Supabase · Vercel

A full-stack concert tracking app, designed, built, and deployed solo, from database schema through production hosting on a custom domain.

  • –Email/password login with optional TOTP two-factor, using Supabase/PostgreSQL row-level security to restrict each user's data at the database level
  • –Custom content moderation with automated filtering and an admin review queue, plus database-backed rate limiting, without third-party services
  • –Continuous deployment through Vercel's Git integration, building and deploying on every push to main
Riffstub landing pageRiffstub artist show page for Tame Impala

Set → Playlist

Python · Spotify API · YouTube Data API

A fun side project: paste a YouTube DJ set and get a Spotify playlist of every track in it, with fuzzy matching to find the right version of each song.

  • –Parses tracklists from video descriptions, chapters, or comments, cleaning up timestamps, tags, and promo text
  • –Scores each Spotify match on title and artist similarity, flagging uncertain ones for manual review
  • –Runs as either a CLI or a browser UI, with tests backing the matching logic

Upcoming

Riffstub for Android

View code
In progress · Coming to Google Play

Kotlin · Jetpack Compose · Supabase · Coroutines/Flow

A native Android client for Riffstub, built solo from scratch. It shares the website's backend, so accounts, privacy, and moderation behave the same on both.

  • –MVVM with one ViewModel per screen, type-safe Navigation Compose routes, and a five-tab bottom bar with a separate back stack per tab
  • –Authorization handled entirely by PostgreSQL row-level security and security-definer functions, so no separate mobile API is needed
  • –Ticketmaster, SeatGeek, and Spotify calls proxied through the website's API routes, so no API keys ship in the APK
  • –Feed, follows, private accounts, nearby-show discovery, TOTP sign-in, and in-app report, block, and account deletion

Skills

Languages

PythonSQLJavaScriptTypeScriptC++

AI & Automation

Claude APIPrompt EngineeringAzure Document IntelligencePlaywrightGitHub CopilotOpenAI CodexGemini

Web / Backend

FastAPIREST APIsReactNext.jsStreamlitSupabaseVite

Databases

MySQLPostgreSQLRedisMongoDB

Cloud & DevOps

AzureVercelDockerCI/CD (YAML pipelines)GitDoppler

Spoken Languages

EnglishSpanish (fluent)

Contact

Feel free to reach out. I'm always open to talking about new projects or opportunities.