I built the products shown here from first concept to working software—shaping product direction, architecture, full-stack implementation, integrations, deployment, and production iteration.*
*For SyncMD, I joined an existing product and team as Head of AI for the SyncSense launch, where I developed its medical-data pipeline.
Systems that answer demand, qualify intent, find live availability, complete bookings, and convert every conversation into usable operational data. Across selected deployments, these architectures have processed more than 1.5 million minutes of calls and reduced operating costs by as much as 92%.
Busy clinics lose calls, take too long to book appointments, and lack a unified view of patient-call outcomes.
Designed and built the system end to end, from AI reception and clinic-specific call logic to protected booking, reporting, and production deployment.
Average booking time fell from 8 to 3 minutes while the measured missed-call rate fell from 12% to 0%.
High-volume customer communication is expensive, inconsistent, and difficult to scale without losing quality.
Defined the product and built the platform from scratch, including multi-tenant voice campaigns, agent logic, CRM tools, scheduling, analytics, and deployment safeguards.
A production platform for operating voice AI across customers, campaigns, calendars, booking systems, and live analytics.
Consumer assistants are good at reminders and chat, but weak at actually completing messy real-world calling tasks.
Shaped the product and built it from scratch: the application, AI task logic, reusable personal context, privacy controls, and voice execution layer.
A product laboratory showing how production voice-AI execution patterns can become a privacy-aware consumer assistant.
Service leads arrive with incomplete context and decay quickly while staff manually qualify and schedule them.
Designed and built the product from scratch, connecting messaging, qualification, ServiceTitan availability, durable booking state, staff tools, and deployment.
A lead-to-booking workflow that responds immediately and carries qualified demand into ServiceTitan.
Two production examples show the path from answered calls to recovered bookings, faster service, and measurable economic value.
AI handles calls that would otherwise be missed and turns qualified demand into booked appointments.
155 recovered bookings per month from AI-handled missed calls.
AI handles calls that would otherwise be missed and turns qualified demand into booked appointments.
603 recovered bookings per month from AI-handled missed calls.
Systems that turn dense medical records and clinical protocols into structured histories, executable study workflows, integrity checks, and evidence-backed budget positions.
Patient histories are fragmented across long PDFs, scans, laboratory reports, visit notes, and duplicated records.
As Head of AI for the SyncSense launch, developed the medical-data pipeline within the existing SyncMD product: staged ingestion, progressive summarization, and source-linked clinical facts.
Thousands of medical pages become a coherent patient history with every conclusion linked back to its source.
Dental teams spend valuable chair time converting spoken findings into formal documentation and charts.
Designed the product and built the Proof of Concept and MVP v1, including voice capture, resilient processing, structured SOAP notes, and periodontal chart data.
Spoken examinations become structured notes and periodontal chart data while the encounter is still fresh.
Clinical study setup turns dense protocols into complex visits, forms, subjects, rules, and review obligations.
Defined and built the product from scratch, covering protocol-assisted setup, eCRF workflows, subject operations, audit trails, integrity checks, and deployment.
One protocol-driven workspace connects study setup, subject execution, data completion, and integrity monitoring.
Research sites must reconcile sponsor budgets with protocol obligations and inconsistent market benchmarks.
Defined and built the full workflow from scratch, connecting protocol extraction, sponsor spreadsheets, procedure mapping, benchmark evidence, and negotiation outputs.
Negotiation teams can identify missing or underpriced work and prepare defensible budget positions faster.
Tools that connect discovery, source analysis, writing, editing, SEO, and publishing into faster, measurable editorial workflows.
Newsrooms need speed, but generic AI writing pipelines lose freshness, editorial context, and accountability.
Shaped and built the editorial system from scratch, including discovery, clustering, writing, review, SEO, provider routing, cost telemetry, export, and video workflows.
An editorial operating system that moves a developing story from discovery through writing, SEO, translation, and export.
Scientific and medical chapters need more than a rewrite: editors have to diagnose structure, verify risky claims, clean language, and preserve meaning across long manuscripts.
Defined and built the right-rail AI review system: diagnostics, structure mapping, fact-check triage, LanguageTool spellcheck, clarity review, formatting, emphasis, visual suggestions, and block-level edit controls.
Editors get a chapter-level critique, a structural map, a queue of risky claims to verify, deterministic language fixes, and targeted rewrites for weak passages instead of a black-box rewrite.
Products that bring conversational AI into purchasing decisions and put AI agents inside the same operating systems as people.
QR menus expose a catalog, but guests still need help deciding, and staff absorb the same questions at the table.
Built an ordering path where guests can ask or browse, turn a preference into an order, and pay without making the floor a FAQ desk.
Guests get answers and a continuous ticket through payment and pickup; the cafe keeps one menu and kitchen status on the same order.
AI agents often operate outside the same tasks, documents, and accountability systems used by people.
Defined and built the product from scratch: the web application, API, agent access, tasks, documents, comments, permissions, and activity history.
People and agents can work from the same source of truth and report progress in the same workspace.