Hi, I am
Sarjan Patel

AI-Powered Senior-Care Automation Platform

• Built AI-driven caregiver automation system using Supabase Edge Functions, Vapi, and Twilio, enabling automated calls and texts for reminders and check-ins with <250ms latency. Designed post-call workflows that summarize conversations and notify caregivers in real time, integrated with Stripe for billing and subscription management.
• Developed and deployed a real-time transcription and reporting pipeline for doctor visits, using OpenAI Whisper hosted on the cloud for HIPAA-compliant speech-to-text processing. Applied NLP models to generate structured medical summaries in caregiver format with ≈400 ms processing latency and high contextual accuracy.
• Built a custom GPT-powered analytics and agent orchestration module leveraging Retrieval-Augmented Generation (RAG) and a dynamic context layer for personalized responses. Implemented an AI agent orchestrator to manage memory, intent routing, and tool execution, enabling adaptive follow-up questions and real-time insights via a dashboard with sub-300ms inference latency.

Unified data platform connecting films, creators, brands, and audiences for cultural impact.

• Architected the full-stack infrastructure for Helix Hub, integrating frontend (Next.js) with backend microservices (Node.js/NestJS) and data pipelines (PostgreSQL, AWS S3, Cloudflare R2). Designed modular content layers for short-form, long-form, and community pages, achieving sub-200ms response latency and seamless synchronization across CMS, analytics, and media services.
• Developed the real-time analytics and event tracking system, implementing distributed data pipelines that capture user interactions across content types. Integrated custom tracking APIs, and a personalized recommendation engine, enabling dynamic audience segmentation and behavior-based content optimization.
• Engineered the orchestration layer connecting CMS, recommendation engine, and video streaming platform, enabling consistent data flow between user engagement metrics, content updates, and recommendation models. Optimized schema design and API routing for scalable performance under concurrent traffic, ensuring 99.9% uptime and reliable end-to-end analytics visibility.

AI-Powered School Commute Platform - Smart Mobility & AI Automation | Built using Flutter, Firebase, AWS, Google Maps API, Cloud Functions

• Architected and developed The Cue App, a cross-platform solution enabling real-time school pickup wait-time estimation through a geofencing-based tracking engine. Designed the full system architecture with live queue visualization, push notifications, and a real-time analytics dashboard, achieving <200ms latency and compliance with COPPA/FERPA standards.
• Developed trip-matching logic for the Parent App, creating a real-time matching engine that connects families with similar routes and schedules. Implemented distance-time-based filtering algorithms and event-driven workflows through Firebase Functions and REST APIs, designed for scalability and future AI-based route optimization.
• Integrated backend systems across Firebase, AWS, and GoTogether APIs, implementing geospatial mapping, distributed data pipelines, and optimized schema design for low-latency queries. Delivered a scalable backend supporting thousands of concurrent users during live school pilot deployments.

Applied AI & Agent Automation Consulting Company

• Project 1: AI Coding platform
  • Helped build the agent-orchestration core. Embedded the new OpenAI Agents SDK to spin up specialized AI agents (Project Manager, Experience Designer, Solution Architect, etc.), enabling non-technical users to create full AI apps through natural-language chat; cut prototype time from days to <10 minutes.
  • Implemented a custom authenticated MCP (Message Control Protocol) Server for Slack and Clickup to enable applications to tap into these services.
  • Unified OpenAI GPT-o3 and Anthropic Claude 4 behind a cost-aware LLM layer, accurately tracking token usage costs.
• Project 2: Medical Documentation AI App
  • OCHO (Organization for Community Health Outreach) is a non-profit that sends medical teams to remote areas to Honduras to offer medical assistance.
  • Engineered a responsive web app that transcribes multi-lingual clinician speech into AI-generated APSO notes (Whisper + GPT-o4), slashing documentation time by 60% during rural brigades in Honduras. Ensured AI generated plan aligned with available medical facilities and treatments in Honduras.
  • Integrated one-click Spanish translation and copy-to-clipboard export, delivering > 95 % accuracy and seamless health record hand-off while preserving medical terminology.

Tempe AZ

• Served as a Judge at Changemaker Central, evaluating student-led venture pitches for impact and feasibility while managing the Farmers Market event end-to-end, including logistics, vendor coordination, member outreach, and on-site operations to ensure smooth execution and strong community engagement.