Jeffrey Sturm

Technical PM who builds — B2B fintech, payments orchestration, agentic AI development, and production-grade web apps

📍 Remote (US) |✉ jssturm@gmail.com |LinkedIn |GitHub

About

Enterprise-grade Senior Product Manager specializing in the intersection of B2B fintech, embedded payments orchestration, and complex ERP integrations (Sage, Acumatica, Dynamics 365 Business Central, Yardi and CDK Fortellis). A dual-threat product leader who combines deep backend API architecture strategies with hands-on frontend execution. Pioneer of agentic AI workflows and vibe-coding methodologies, utilizing local development setups (Claude Code, VS Code, MCP servers) to personally build, deploy, and validate functional UI/UX mockups. Proven track record of accelerating product velocity, decoupling complex legacy infrastructures, and optimizing financial interfaces for maximum user flow efficiency.

Core Competencies

0→1 product AI assisted development roadmap strategy user research competitive analysis A/B testing stakeholder alignment Node.js TypeScript SQLite Playwright Docker API design payment orchestration ERP integrations B2B payments accounts payable automation accounts receivable automation virtual cards ACH processing supplier/vendor management PCI compliance Cross-functional team leadership engineering-product bridge builder culture advocacy

Experience

REPAY — Realtime Electronic Payments
Senior Product Manager / Senior Product Owner / UI/UX Design
May 2025 – Present
  • AI-Enabled Delivery Velocity & Frameworks: Institutionalized an organization-wide agentic AI framework leveraging Claude Code and custom Jira/GitHub MCP integrations. Developed comprehensive default protocols for AI agents to enforce engineering efficiency and logic validation during product development.
  • Hands-on Functional Prototyping: Embedded AI-generated functional prototypes directly into Epics and User Stories, utilizing AI tools to bridge the gap between product requirements and engineering pods, instantly aligning distributed teams and slashing development rework.
  • Ecosystem Experience Ownership: Directly steer the end-to-end digital experience across the embedded payments and ERP integration ecosystem — buyer, vendor, and partner surfaces.
  • Enterprise PropTech & ERP Architecture: Manage complex core integrations for 15+ leading enterprise platforms—specifically including real estate/property management systems like Yardi alongside Sage and Dynamics 365—directly driving ecosystem adoption.
  • User Experience (UX) Optimization: Partner daily with cross-functional Engineering teams to translate product vision into intuitive interfaces; optimized dashboard layouts by strategically modifying feature alignment to reduce user friction and accelerate interaction velocity.
  • Platform Modernization: Architected a near-real-time (NRT) orchestration gateway to decouple deployment cycles, enabling rapid feature iteration and streamlined transaction tracking across heavily fragmented software landscapes.
FIS (Fidelity National Information Services)
Digital Product Manager
July 2023 – Feb 2025
  • Multi-Surface Platform Ownership: Owned and optimized the digital end-user experience across three distinct ecosystem surfaces—the Customer Site, Vendor Site, and Partner Site—ensuring seamless user journeys and cohesive interface design across all multi-tenant portals.
  • Award-Winning Product Execution: Served as Primary Digital Product Manager for the platform awarded the "Best Business Payments Solution (2024)", successfully scaling the division through a period of 12% average YoY growth.
  • Cross-Functional Team Empowerment: Acted as the core technical SME to empower 15 distributed engineering teams and 8 enterprise sales teams, providing technical insights that directly closed 30 net-new enterprise deals.
  • Agile Transformation: Architected an optimized Agile planning model that slashed planning timeline overhead by 80%, compressing delivery planning cycles from 2.5 days down to 0.5 days.
FIS (Fidelity National Information Services)
Director, Product Operations & Solutions Consultant
Jan 2022 – July 2023
  • API & Core Modernization: Led a multi-year infrastructure transformation, successfully migrating legacy SOAP APIs to high-performance RESTful web services for 80+ white-labeled client sites.
  • Automation & Experience Velocity: Revolutionized automated transaction processing mechanics, triggering a 30% reduction in processing velocity times and a 25% increase in system transactional capacity.
  • P&L and Vendor Management: Captured a 15% reduction in operational spend by aggressively renegotiating critical third-party vendor contracts, removing $2.2M in redundant licensing overhead.
FIS (Fidelity National Information Services)
Product Operations Manager (ACH & RTP)
July 2021 – Jan 2022
  • Core Clearing Operations: Established foundational Product Operations frameworks for core real-time clearing and automated business units, dropping transaction processing errors by 30%.
  • Accelerated Time-to-Market: Engineered a 20% acceleration in time-to-market delivery by authoring comprehensive PRDs that elevated client satisfaction metrics by 40%.
Prior Professional Experience Summary
  • Lead Project Manager / Associate Product Manager (FIS): Managed enterprise software integrations with a 95% on-time completion rate; boosted profitability by 35% via strict scope and cost-tracking.
  • Technical Project Manager (Allstate / Keefe Group / ComNet): Orchestrated a portfolio of 50+ concurrent infrastructure projects scaling to $10M annually; compressed QA/testing cycles by 30%.
Product Engineering & Architecture
  • Node.js v24 runtime for payment orchestration services
  • SQLite persistent cache engine for local-first data integrity
  • Playwright for browser-based automation, testing, and layout validation
  • Docker containerization for consistent development and deployment environments
  • Built Vendor and Vendor Admin sites providing awareness of inbound payments and visibility into expiring virtual cards
  • Designed modular payment plugin architecture keeping payment components strictly decoupled from core AP/AR ledgers
  • Built dashboards and monitoring surfaces for buyer AP managers, controllers, and CFOs

🚀 Projects Built with Agentic AI

These production-grade applications were independently designed, engineered, and deployed from scratch using agentic AI development workflows — demonstrating the ability to ship complete, polished products without a traditional engineering team.

ATHENA AUREUS — Decision intelligence for humans in charge
High-stakes decisions enterprises can explain, challenge, and measure — live at athenaaureus.com
July 2026 – Present
  • What it is — Public product for recording and improving how an enterprise decides. Built for high-stakes calls where accountability matters; recommendations are designed to be checked later against outcomes. Work happens in Athena Aureus Desk (the workspace). Site: https://athenaaureus.com/.
  • What it is not — Not an “AI employee,” not autopilot that replaces people, not legal eDiscovery, and not Amazon Athena (the SQL analytics product — same first word, different tool).
  • Beyond chatbot answers — Generic chat gives confident advice that disappears when the tab closes. Athena Aureus turns the organization’s highest-stakes calls into a repeatable decision process that gets sharper over time.
  • Explain · challenge · measure — Recommendations must be explainable (evidence on the record). Challenge is invited before confidence hardens. Predicted risk and results are compared to what actually happened so judgment quality compounds.
  • Humans stay in control — The product can recommend; a named person must confirm before anything consequential ships. Leaders can override with their reason attached — accountability stays with people.
  • One governed path, durable memory — Releases, vendors, capital, and exceptions can run the same path. Who decided, on what evidence, under which priorities, who objected, what was promised, and how it turned out stays searchable after meetings and chat threads are gone.
Trading Bot — Automated Multi-Strategy Trading Engine
Paper trading bot with dual-strategy engine, risk management, and n8n health monitoring
June 2026 – Present
  • Dual-Strategy Engine — Built a TypeScript trading bot in Node.js 24 that monitors 29 symbols (17 equities + 12 crypto) and generates BUY/SELL/SHORT/COVER signals via two complementary strategies: a conservative TrendFollowing strategy (EMA crossover with RSI/volume confirmation) and an aggressive MomentumAggressive strategy (breakout momentum with pyramiding). Crypto trading supports LONG/SHORT dual-direction signals.
  • 8 Hard Risk Constraints — Daily loss limit, weekly loss limit, max drawdown, position sizing, portfolio risk cap, cash reserve floor, VIX circuit breaker, and a paper-only gate that makes live orders impossible when enabled. Real-money trading requires an explicit multi-step opt-in.
  • Real-Time Market Data — Quotes, 5-minute bars, and daily bars via market data APIs. PM2 process manager for persistent execution with automatic restart on failure.
  • n8n Health Monitor — Designed and deployed a self-hosted n8n workflow automation instance with a dedicated health monitor workflow. Every 5 minutes, the workflow pings the trading bot's health endpoint, evaluates the response, and triggers alert logic on failure — closing the observability loop for an unattended automated system.
  • Docker Compose Deployment — Containerized the full stack (trading bot + n8n + health monitor) for consistent development and deployment environments.
Elevated-Applicant — AI-Powered Job Search Command Center
Full-stack, local-first web application (with an optional CLI) automating every phase of the job hunt
March 2026 – Present
  • Web Dashboard (Next.js 16, 17 pages) — Default local experience after install: pipeline kanban, AI-powered job explorer, interactive CV builder with 20 industry-specific templates, company intelligence with automatic classification (enterprise/startup/scale-up), interview prep, keyword optimization, and DeepSeek-powered AI chat assistant. Free install builds the app locally; run elevated-applicant serve → http://127.0.0.1:3000. CLI / AI coding assistant optional.
  • Verified local install — One-command installer ships a checksum-verified release artifact, installs deps, builds the production web UI, and sets up Playwright. Updates preserve user data. Node.js 20+.
  • Discovery Engine — 58 provider modules scanning 78+ job boards and public ATS APIs (Greenhouse, Lever, Ashby, Workday, BambooHR, and 53+ more). Three web search providers (Google Jobs, DuckDuckGo, SearXNG) discover jobs from Indeed, LinkedIn, Glassdoor, and other aggregator sites with zero API keys — queries auto-adapt to your target roles from profile config. 5 reverse-scan sources programmatically walk every known company board.
  • AI-Agnostic Evaluation — 8-block structured scoring (role summary, CV match, compensation research, STAR-based interview prep, posting-legitimacy check, red flags, offer negotiation). 33+ specialized prompt modes. Model-agnostic core runs on 5 backends: Gemini (free tier), Ollama (fully local), OpenAI, DeepSeek, and OpenRouter (multi-model). Built-in DeepSeek chat assistant for career coaching.
  • Document Generation — ATS-optimized PDF CVs via Playwright headless Chromium, LaTeX typeset CVs, keyword-mirrored cover letters, recruiter-ready one-pagers, and recruiter packs — all from a canonical cv.md file.
  • Pipeline Tracking & Analytics — File-first canonical store (human-readable, git-diffable) with SQLite-derived index. 5 tracker scripts for atomic operations. Go + Bubble Tea terminal dashboard with 6 filter tabs, 4 sort modes. SQLite analytics for response rates, salary trends, and portal effectiveness.
  • CI/CD & Quality Infrastructure — 500+ test checks. 5 GitHub Actions workflows: type-check + web build on every PR, E2E Playwright smoke tests across 14 pages, Docker build verification, CodeQL security analysis, and dependency review. Self-update system with strict file boundary enforcement. Docker Compose deployment.
  • Multi-CLI & Privacy — Works natively with Claude Code, Codex, OpenCode, Gemini, Qwen, Grok, and Antigravity. English & Spanish documentation. Local-first — no cloud middleware, no data collection. Your API keys, your machine.
Plan-It — AI-Powered Itinerary Planner
Self-contained travel planning — minute-by-minute itineraries for any venue type
January 2026 – March 2026
  • Cross-Platform SDK & Installer — Pip-installable package (pip install plan-it) with cross-platform installer script supporting macOS, Linux, WSL, Windows, Termux, and iSH. Features --dry-run environment validation, 3x retry with exponential backoff on downloads, disk space checks, and PyPI → local source → GitHub tarball fallback chain. Fully self-contained — zero API keys required, works OOTB with DuckDuckGo.
  • Itinerary Engine — Multi-stage natural-language intent parser with 40+ known venue aliases and city-qualified disambiguation. Auto-classifies destinations into 7 venue types (theme parks, museums, zoos, national parks, festivals, city tours, general) each with tailored opening hours, attraction counts, and departure times. Priority-tiered scheduling with dynamic wait-time scaling by time of day. Multi-day trip detection via text signals + drive-distance threshold
  • Curated Knowledge Base — 40+ pre-vetted venues with specific ride and exhibit names: every major Disney/Universal park, SeaWorld, Busch Gardens Tampa & Williamsburg, Cedar Point, Six Flags Over Georgia, Kennedy Space Center, The Louvre, The Met, Smithsonian, San Diego & Bronx Zoos, 8 national parks. Multi-tiered matching with 5 fallback strategies
  • Web Research Engine — Zero API key DuckDuckGo client with rate-limited queries. 6-query venue research pipeline. Restaurant intelligence with 15+ garbage-pattern filters and 18 cuisine types. Hotel search with 70+ chain validation and 6-city curated fallback database. Dual-layer geocoding (US Census Bureau + Nominatim/OSM). OSRM real road-network routing for actual drive times. 50-state name/code normalization
  • Traffic & Alert Intelligence — Pattern-matched corridor warnings for I-4 Orlando, I-95 Miami/Jacksonville, Atlanta metro I-75/I-85, and Tampa Bay I-275 with rush-hour estimates. Case-insensitive alert dedup. Midpoint city inference for 10 common route pairs with automatic rest stop insertion
  • Full-Stack Architecture — FastAPI backend with 10+ Pydantic models and OpenAPI docs. 4-operation PATCH-based schedule editing. In-memory plan store with UUID-keyed retrieval. Per-IP rate limiting and constant-time API key auth. Vanilla JS SPA frontend (zero build step) with inline result preview, modal-based schedule editing, inline reminder system, sessionStorage persistence, and AM/PM departure time toggle. Cross-platform pip package with launchd/systemd/Scheduled Task startup
  • Optional DeepSeek AI — Enhanced planning via DeepSeek LLM (deepseek-v4-flash, deepseek-reasoner, deepseek-coder). Multi-agent framework for specialized planning tasks

Let's Connect

I'm always open to discussing new opportunities, collaborations, and interesting problems.

✉ jssturm@gmail.com · LinkedIn