Michael Alexander

[email protected] | linkedin.com/in/michael-alexander-nj | South Jersey / Philadelphia Metro

Choose Your Format

Same career, two ways to read it. Pick whichever fits how you review candidates.

The Full Story Long-form narrative — project deep-dives, the decisions behind the numbers, and how I actually think. Best if you want to understand who you'd be hiring. PDF · ~5 pages Download ↓ Recruiter / Quick-Scan Concise and traditional — scannable bullets, standard sections, ATS-friendly. Best for recruiters and applicant-tracking systems. PDF · 3 pages Download ↓

About

I'm a technology leader and hands-on operator. From 1998 to 2025 I ran the technology function at a multi-channel e-commerce business that grew from a startup to $8M in peak revenue — infrastructure, identity, networks, data, integration, end-user systems, crisis response, and a team of 10+. Since 2025 I've been building production AI-powered SaaS tools and treating Claude Code as my primary development environment across code, infrastructure, and prototypes.

The thing I bring to a team is a particular kind of operator instinct: I close the gap between what someone reported, what the system is actually doing, and what the business actually needs. I fix root causes rather than symptoms. I absorb operational complexity so leadership can focus on strategy. And I don't reach for AI as a blanket solution — LLMs are a tool, applied where they create genuine value, with traditional logic everywhere else.

Three decades of multi-system operational fluency. Specific tool names matter less than the meta-skill: when a new system shows up in the operation, I learn it because that's the work, not because it's a project.

What I'm Building (2025 – Present)

Independent Consultant / Founder — AI/SaaS Development

2025–Present

Building production AI-powered SaaS tools end-to-end. Problem-first: every project starts with a real problem, and AI is one instrument in the toolbox — applied where it creates genuine value, left out where it doesn't.

  • Real-Estate SaaS Rebuild — 3 days, spec to working application. Authored a detailed specification, handed it to Claude Code, and got a one-shot rebuild that recovered all prior functionality and most new features; a few focused sessions finished the rest. That same scope was a roughly six-month solo project two years ago — the productivity delta when disciplined spec-writing meets agent orchestration.
  • AI Content Scoring Platform. Multi-LLM provider architecture across OpenAI, Anthropic, Google Gemini, and local quantized models — cost-optimized routing, automatic fallback, production-grade bulk API. A 6-model bake-off on real workloads put Gemini 2.5 Lite ahead on the cost-quality envelope for text analysis, beating the "always reach for the frontier model" default.
  • Multi-Agent Document Processing Pipeline. Classifies inputs, routes them to specialized agents on extracted criteria, and enforces quality at agent boundaries (verbatim source preservation, no-fabrication rules, traceable provenance); persistent state, 100+ item batches, human-in-the-loop checkpoints, cross-run memory.
  • "Open Mic" adaptive lyric scroller (iOS, in progress). SwiftUI app that listens to live performance audio, detects tempo in real-time, and adjusts scrolling speed — and trains to an individual's timing across sessions. Real-time audio analysis, on-device ML, adaptive UX.
  • Reddit Signal Scanner (production). Multi-stage pipeline scanning target subreddits for product-opportunity signals: selection → thread analysis → cross-thread pattern detection for emerging complaint and need clusters.
  • ADHD-focused task manager (iOS, shelved with intent). Built core functionality to feature-complete, then shelved it rather than ship a watered-down version when iOS constraints eroded the ADHD-specific value. When the workaround kills the value prop, stop building.
  • Infrastructure. 3-host Docker home lab running 6 application stacks on a Jenkins-driven multi-environment CI/CD pipeline, Cloudflare Tunnels, and Proxmox across mixed hardware — backups, monitoring, patching, and controlled change applied to a system I built and maintain myself.
  • Tool & model evaluation as practice. A 4-month side-by-side trial of GitHub CoPilot vs Claude Code; also evaluated Cursor, Cline, and Gemini's tools — chose Claude Code on output, not allegiance.

What I Did Before (1998 – 2025)

Technical Operations Leader — Nella Cosa Corporation

Marlton, NJ 1998–2025

Multi-channel e-commerce · $8M peak revenue · Team of 10+ · 100+ concurrent system users · Complete technology operations ownership from startup to enterprise.

  • Built the omnichannel platform that captured 60% of revenue. Architected custom commerce integration across Amazon, eBay, Walmart, and direct channels before commercial tools existed — 0% to 60% of total revenue ($4.8M+) within 3 years, syncing inventory and channel-specific data across 20,000–25,000 active SKUs where a single base product could explode into 500+ variants.
  • Re-architected the company's barcode system under an Amazon GS1 mandate. Licensed a GS1 prefix and built a pool table as single source of truth — assignment, retirement, and cooldown-based recycling — then scripted propagation into inventory and bulk SQL-to-PDF label generation. After a vendor label-scan failure surfaced via an Amazon complaint, I moved QA to the choke point where the work changes hands: a spot-check at container check-in.
  • Cut fabric waste on our highest-volume product without trading it for inventory waste. The obvious fix — printing slower-selling sizes in the leftover strip — would have moved waste from the roll to the shelf. I treated it as a layout problem and designed mixed-size print layouts matched to demand ratios, not just the fabric.
  • Eliminated daily system outages. A SQL Server migration and infrastructure modernization cut outages from 2/day to 1 every few months (99.9% improvement) and unlocked previously impossible reporting — while I held a 3–6 ticket/day support load as the user count grew 100x, by fixing root causes, not stacking workarounds.
  • Held the line through a $1M+ COVID surge. When bandanas were named as a mask substitute, the site took $500K on Amazon in 15 minutes. As sole technical responder, I wrote new code and processes over 48 hours to prevent failure at 10x volume — zero downtime. Those processes stayed in production 4+ years.
  • Drove 15–25% revenue uplift ($1.2–2M on an $8M base) through pricing optimization and business intelligence; 20% YoY growth in the best year; average order value $40 → $50.
  • Operated lean. Ran the entire technical infrastructure on a ~$100K annual budget. Consolidated 3 physical servers into 1 (70% power reduction), saved $17K going refurbished on a single purchase, cut shipping errors 5% → 2%, and scaled to 500 orders/day peak.
  • Led 10+ people. Built training programs, operational procedures, and a culture of ownership; translated complex technical concepts into executive-ready language across ownership, sales, operations, fulfillment, and finance.
  • Extended the ERP for 25 years. Wrote integration code, custom queries, automation modules, and reporting logic against a live production Microsoft Access + SQL Server data layer — covering operational needs commercial software didn't address.

Core Competencies

Tool Fluency — Three-Decade Breadth

  • E-commerce platforms (operator-level): Amazon Seller Central (Expert), eBay, Walmart Marketplace, Miva — and the broader retail platform landscape across 25+ years
  • Order / shipping systems: UPS WorldShip, FedEx Ship Manager, USPS Endicia, ShipStation — and most major commercial shipping integrations of the past two decades
  • Productivity / collaboration: Microsoft Office (every major version since Office 95), Google Workspace, M365
  • ERP / business systems: custom-extended Access + SQL Server ERP (25 years), QuickBooks, integration-side exposure to enterprise ERPs

Strategy & Leadership

Technology Strategy · Digital Transformation · Change Management · P&L Responsibility · People Management (10+ direct reports) · Crisis Management · Stakeholder Communication

Infrastructure & Systems

Active Directory and IAM · Windows Server (on-prem and cloud) · LAN/WAN networks (multi-site WiFi, hands-on Cat6) · Backup and patch management · Hyper-V / Proxmox virtualization · Docker / Docker Compose · Cloudflare Tunnels · Jenkins-driven multi-environment CI/CD

Data, APIs, and Development

Microsoft SQL Server (Advanced — 15 years) · MySQL · PowerShell (Advanced — 10 years) · Python · Swift / SwiftUI · REST API design · Webhook integration · Master-data design and single-source-of-truth modeling · Real-time audio analysis

AI / LLM Implementation

Multi-provider architecture (OpenAI, Anthropic, Google Gemini, local quantized models) — cost-optimized routing, automatic fallback, empirical head-to-head model evaluation. Agent orchestration with Claude Code. Context and token engineering. N8N workflow automation. Working familiarity with RAG and Model Context Protocol — applied project-by-project after measuring cost/benefit.

Education

Rutgers University — Camden

Dual Focus: Physics and Computer Science — a foundation in analytical reasoning, hypothesis-driven methodology, and systems modeling.

A Few Things That Aren't on the Resume

I love taking things apart to see how they work, then asking whether they can be better — from electronics to software to AI systems. Outside of work I spend time with my wife Eileen and our rescue greyhound Maggie. I'm a live music regular, a photographer, a gamer, a gardener, and a hiker. I came of age when learning a new system meant buying a book at Barnes & Noble; I've never stopped learning new systems, and a Claude-Code-augmented ramp on a new tool is faster than reading-a-book ramp ever was.

What I'm Looking For

Problem-solving and innovation work, hands-on but strategic — places where deep operational expertise and current AI/LLM fluency both create value. Open to remote and reasonable hybrid arrangements (≤60 min commute for 1–2 days/week, ≤30 min for 3+ days/week). Available immediately.

If any of this resonates, I'm easy to reach: [email protected].