Charlie M.

I help businesses deploy AI into their day-to-day operations so the same team gets more done with fewer handoffs — and I build the systems of record that make that possible for document-heavy firms in legal, lending, credentialing and healthcare, and for research teams whose platforms must hold up under federal grant review. I scope the work, make the architecture calls, and answer for the result. AI IN BUSINESS OPERATIONS AI changes what a small operations team can carry. Documents arrive from various sources (email, scans, portals), data lives in three systems that don't agree, and someone re-keys it. Reading a document, classifying it, pulling the fields out, deciding where it goes, reconciling it against the record in another system: that is work people used to do by hand, and it is now work a model does in seconds with a person checking the exceptions. My job is deciding where a model belongs in your workflow, where it doesn't, and what check keeps it accurate. I treat the models themselves (OpenAI, Anthropic Claude, Google Gemini, connected through MCP to your CRM and document systems) as interchangeable; the value is in the integration, the human-in-the-loop design, and the monitoring. What that has looked like in practice: OCR and document AI turning handwritten rental applications into clean Salesforce records for a property manager; LLM orchestration that lets a research team compare models and prompting strategies in production; HubSpot and Zapier automations rebuilt so a solar company's revenue ops and project cost audits run on numbers finance can trust. OPERATIONS PLATFORMS When off-the-shelf tools stop fitting, I build custom CRMs and proprietary internal operations platforms — the system of record a firm actually runs on — because owning the data model is what gives you flexibility now and room to scale later, and it is what lets AI plug into the workflow instead of bolting on beside it. On a lender's platform the decisions were structural: an Azure migration that met financial-workflow security expectations, and a due-diligence integration layer built for retries, error states and traceability when third-party APIs fail. Same job on a credentialing portal's Salesforce API layer and a surgical practice's analytics pipeline. RESEARCH SOFTWARE As technical co-founder on two grant-funded mathematics platforms, I built the prototype behind a Phase I award, then designed the Phase II architecture the $1M federal SBIR award depended on: a React/Firebase client orchestrating an AWS-hosted AI service, FERPA-compliant logging, and reporting built for principal investigators. On a foundation-funded study I shipped the MVP in under 30 days. I also provide the technical build for an NSF CAREER-funded AI-driven education study platform. TECH BACKGROUND Expert-Vetted on Upwork (the top 1% of talent, vetted by Upwork's own process), with 7+ years building software professionally. PhD from Northeastern; senior software engineer at HubSpot, working inside the CRM platform I now integrate with for clients; lead product developer at AppGeo. Founded Smplfy in 2021. Day-to-day stack: NestJS, React, TypeScript and Node on the application side; Java and PHP where clients already run them; MySQL, Redis and Firebase for data; AWS and Azure for infrastructure; OpenAI, Anthropic Claude and Google Gemini through direct APIs and MCP for the AI layer; Salesforce, HubSpot, Zapier and Stripe as the systems most of that work plugs into. HOW I WORK Discovery and architecture are mine: one call, then a written scope naming the systems involved, where AI actually earns its place, and the risks. When the work outgrows one engineer, I bring in my Smplfy team and stay your single point of accountability. For multi-year platform ownership, start at the Smplfy agency profile.