A little about me

Hi, I’m Jai.

I build consumer products and the systems behind them. Right now, I’m the co-founder of Vora, helping people notice useful things about their health without having to ask.

Before that, I was on the core team at AWS SageMaker. I studied at Penn in the Management & Technology program.

Work

VoraCo-founder · Current

We’ve built proactive AI around workouts, nutrition, sleep, and recovery, with wearable integrations including Oura and Garmin. Now we’re developing a new experience for bringing those insights to people.

An earlier chapter

Ziffyn

An at-home chef service, built across three connected apps.

I taught myself iOS development and built three connected applications from scratch. The challenge was making the customer, marketplace, and chef/driver experiences work together as one service.

MarketplaceChef & driverConsumer

Read the Ziffyn case study ↗

Amazon Web ServicesPreviously

I contributed to SageMaker inference capabilities for running multiple foundation models on shared endpoints, with separate resource allocation and scaling. I also contributed to asynchronous model invocation in Amazon Bedrock.

SageMaker launch Bedrock API

Education

University of Pennsylvania

Management & Technology program.

Selected code

From operating systems in C to product-data pipelines and consumer apps. These are historical projects; each repository documents its setup, collaborators, and limitations.

PennOS · An operating system in C

I helped build a Unix-like operating system for a Penn team project: a preemptive scheduler, process management, a shell, and a FAT-style filesystem. It runs in Linux userspace, using timer signals and saved execution contexts to switch between processes.

C · Penn team project Source & architecture ↗Scheduler ↗

PennShell · Inside a Unix shell

Mytreyi Narayanaswamy and I pair-programmed a shell in C, contributing roughly equally. It handles pipelines, I/O redirection, signals, and foreground/background jobs, including the process and terminal control behind jobs, bg, and fg.

C · Pair-programmed Source & examples ↗

Making sense of messy product listings

At Auricore, I built pipelines to collect listings from luxury resale sites, categorize products, process images, and store catalog records. A second pipeline reconciles inconsistent listings using source-quality checks, AI attribute extraction, visual search, and human review of duplicates.

Python · DynamoDB · S3 Ingestion ↗Canonicalization ↗

From a photo to an item valuation

At Auricore, I worked on identifying items from photos, researching market prices, and connecting the results to borrower and lender applications. The project spans a web app, supporting APIs, and payment integrations.

Next.js · TypeScript · Firebase Web source ↗Services ↗

Auricore on mobile

A companion app with guided item intake, category-specific AI conversations, inventory, and payment onboarding.

React Native · TypeScript Mobile source ↗

Other work

Fine-wine traceability

I architected and developed a supply-chain platform connecting a custom blockchain network API with IoT trackers for temperature, humidity, and light exposure, from vineyard to cellar.

I’m also part of Tréfo and Open-UAV, groups focused on agricultural vision and drone simulation. Research and team-project notes ↗

Poems

I love to write. I’ll be sharing some of my poems here.

Places I’ve been

I love to travel. Here are the countries and territories I’ve visited so far.

Explore the mapSelect a highlighted country or a name below.

VisitedMap data: Natural Earth

Peru · The Amazon and Machu Picchu.

North America & Caribbean

South America

Europe

Africa

Middle East

Asia

A note on the theme

The Heroes of Might and Magic theme is for my grandfather, an avid gamer back in the ’80s and ’90s. It’s a small tribute to him and his love of those early games.

For my grandfather.