Manav Patel

• Developed and deployed end-to-end AI-driven audio processing pipelines (MERN + Python) for top-tier U.S. insurance clients, powering call-center analytics and liability claim automation
• Architected asynchronous, event-driven systems using advanced hashing, cutting processing latency by 10.1x
• Built a predictive analytics module to surface high-value stakeholders, driving a 7x increase in B2B onboarding conversion
• Championed CI/CD automation, code review, and observability practices across the engineering team

Curator's note — The artist's audio-AI period. This website's previous design was a call transcript — the curator has since rotated the collection.

• Led full-cycle development of multiple AI-integrated web applications, from requirements through production deployment
• Built scalable MERN full-stack systems embedding recommendation engines, audio classifiers, and behavior analytics
• Designed cost-efficient, AI-ready AWS infrastructure (Lambda, S3, EC2), cutting compute utilization by 52%

A FastAPI backend that lets users upload text and PDF radiology reports, automatically processes them via OpenAI's API for analysis, and stores results in DynamoDB Local with MinIO-backed file storage and webhook-triggered processing.

On view: Backend ↗ · On view: Frontend ↗

An AI-powered elder-care platform tackling loneliness among aging populations, delivering real-time health analytics and predictions through an intelligent care support assistant.

On view: Repo ↗

Curator's note — Completed in 36 hours at Stanford TreeHacks. The medium is adrenaline.

An end-to-end pipeline that ingests audio, auto-detects and censors sensitive personal information (names, SSNs), and layers on speaker identification and transcript generation.

On view: Repo ↗

Curator's note — A study in discretion — the machine listens carefully, then forgets what it should not repeat.