Skip to content

Archit MehtaFounder & engineer.

I’m a tech entrepreneur building Stamp, the AI Secretary. Previously a software engineer at Stripe and Apple, and President of Cornell AppDev.

See my workGet in touch

archit@mehta: ~ — zsh

    archit@mehta:~$
    Tab autocomplete·↑ history
    Archit Mehta smiling, wearing a black hoodie.

    Archit Mehta

    CEO of Stamp (YC W25)

    stampmail.ai
    • San Francisco, CA
    • CS + Applied Math, Cornell

    Hi, I’m Archit. Builder, founder, engineer.

    I’m a tech entrepreneur currently building Stamp, the AI Secretary. Stamp is a new email platform that provides every user with their own personal Secretary that triages, writes, and learns like them. Stamp is also backed by Y Combinator.

    I’ve built dozens of projects with hundreds of thousands of users, and as an avid entrepreneur I’m always tinkering and building new ideas with the potential to revolutionize the world.

    Prior to founding Stamp, I worked as a software engineer at Stripe, where I saved the company more than $1 million annually through the performance improvements I made.

    I graduated from Cornell University with a degree in Computer Science and minor in Applied Mathematics. On campus, I was President of Cornell AppDev, a team of 50 students building apps with over 15,000 combined active users.

    By the numbers

    YC W25
    Stamp, the AI Secretary
    10M+
    emails processed through Stamp
    $1M+
    saved annually at Stripe
    100,000+
    users across my other apps

    Where I’ve been

    • Stamp

      CEO & founder of Stamp (YC W25), the AI Secretary for email.

    • Stripe

      I was previously a Software Engineer at Stripe, saving the firm $1 million annually in database costs.

    • Apple

      I’ve interned at Apple, where I constructed a 10,000+ line C++ library from scratch to aid navigation of autonomous vehicles.

    • Cornell

      I studied CS & Math at Cornell, where I also conducted Computer Vision research.

    • Cornell AppDev

      I was president of Cornell AppDev, a team of 50+ student software engineers building products for 15,000+ users.

    Where I’ve worked.

    From YC-backed founder to engineer at Stripe and Apple: the teams and problems that shaped how I build.

    1. Chief Executive Officer

      Stamp (YC W25)

      Jan 2025toPresent

      • Architected and Implemented a service oriented backend integrating directly with IMAP and SMTP email protocols, supporting all modern email client features (schedule send, snoozed emails, labels, etc)
      • Built an IMAP-optimized Agent SDK to construct a multi-agent ecosystem, with distinct agents for email annotation (summarization, todos, draft replies), user memories, Deep Research, Voice Mode, and more
      • Fine-tuned a multilingual embedding model via contrastive learning to implement both a Retrieval Augmented Generation pipeline (hybrid semantic search + reranking) and a semi-supervised clustering email labeling algorithm
      • Built Web, iOS, Android, and Desktop apps, with on-device persistence, notifications, and background refresh

    2. Software Engineer

      Stripe

      Jul 2024toDec 2024

      • Saved $1 million annually in database costs by optimizing PaymentIntents API code, and performing shard-aware database migrations for 150 billion stale records via PySpark, Databricks, Apache Airflow, and Apache Iceberg
      • Improved API performance and stability for the PaymentIntents Multicapture API by using the Factory Design Pattern to decouple code across customer environments, increasing test coverage by 20%
      • Improved Adaptive Pricing API semantics by adding presentment currency fields to the PaymentIntent Dataview and refactoring legacy PaymentIntent APIs to support multi-currency presentment

    3. Software Engineer Intern

      Apple

      May 2023toAug 2023

      • Implemented custom C++ library to detect topological and geometric changes in HD Maps for autonomous systems, reducing data transition time by 70%
      • Constructed custom UI geometry classes, modals, and event handlers in Objective C to visualize library analysis in custom MacOS application
      • Architected Protobuf schemas for library analysis serialization, visualization, and deployment

    4. Computer Vision Researcher

      Research at Cornell

      Aug 2023toMay 2024

      • Working with Prof. Noah Snavely to improve the accuracy of video entity motion detection pipeline
      • Implementing CoTracker optical flow pipeline for data preprocessing and optimizing neural network efficiency
      • Investigating applications of pipeline's bijective mappings to 3D canonical space in generative AI video editing

    5. President

      Cornell AppDev

      Feb 2022toMay 2024

      • Leading all Software Development (iOS, Android, Backend) for a team of 50 students building apps with 10,000+ active users
      • Constructing team vision, overseeing logistics, and communicating with stakeholders in Cornell administration
      • Previously served as Product Lead, Technical Product Manager, and Backend Developer

    6. Software Engineer Intern

      Johnson & Johnson Robotics

      May 2022toAug 2022

      • Worked on robotic surgery cloud platform enabling real-time data analytics & communication between IoT robot devices
      • Automated deployment of 5,000+ cloud resources (Azure IoT Hub, Blob Storage, etc) by creating Terraform modules integrated into Enterprise Jenkins Instance
      • Conceived & implemented Azure IoT Edge authentication module used across 8 device teams to interface with cloud

    Products I’ve shipped.

    Apps, websites, and models with I've built from zero to production, with hundreds of thousands of users. Click on any card to read the full story.

    More on GitHub
    • Strompt: Structured Prompts screenshot

      Strompt: Structured Prompts

      Strompt (structured prompts) is an opinionated, zero-dependency TypeScript library for building readable, correct, and maintainable prompts for agentic applications. Instead of sprawling template strings, prompts are composed with a fluent builder of sections, subsections, and lists that render to clean Markdown with automatic numbering, making them easy to debug, diff, and reuse. Renderers are fully extensible, so teams can override how any node is rendered or define their own from scratch.

      • TypeScript
      • npm
      • Fluent Builder API
      • Markdown
      • LLM Agents
      npmGitHub

    • Archivist: AI Code Search

      Archivist is a VS Code extension that brings blazingly fast, AI-powered semantic search to any codebase. It indexes a project in minutes using a custom-trained code embedding model, answers natural-language queries in milliseconds, and summarizes matched snippets to explain what they do, keeping its index in sync as you edit. Code is chunked with Tree-sitter syntax trees for precise results, and your source is never stored.

      • TypeScript
      • VS Code API
      • FastAPI
      • Sentence Transformers
      • Tree-sitter
      • Pinecone DB
      • MongoDB
      • Docker
      VS Code MarketplaceGitHub
    • Cornell Course Engine screenshot

      Cornell Course Engine

      Integrated directly with Cornell's class roster, this course search engine utilizes a custom Semantic Indexing and Search pipeline to help Cornell students discover interesting courses based on natural language queries.

      • React
      • Flask
      • HF Transformers
      • Pinecone DB
      • EC2
      WebsiteGitHub
    • Yolo: Social Events screenshot

      Yolo: Social Events

      Yolo is the first social media app built around your calendar. It's designed to help you find fun events on campus and make friends with similar interests, and launched at Cornell in Fall 2022!

      300+ Daily Active Users, 1000+ Users

      • React Native
      • MongoDB
      • Express.js & Node.js
      • Socket.io
      • AWS CloudFront
      • AWS S3
      • AWS EC2
      App StoreGitHub
    • Newsflash: AI News screenshot

      Newsflash: AI News

      NewsFlash is a website and mobile app that utilizes Machine Learning to identify news bias. It displays top headlines (and their bias), and is capable of analyzing the sentiments and bias of any article on the web.

      1,000+ Users; 50,000+ Articles Analyzed

      • TensorFlow
      • NLTK
      • React Native
      • Flask
      • SQL
      WebsiteApp Store
    • Volume: Cornell Publications screenshot

      Volume: Cornell Publications

      Volume is a mobile app which aggregates all student publications, magazines, and flyers on Cornell campus. I led a team of 10 developers, designers, and marketers to build the app and grow it to over 400 users.

      400+ users; 20+ publications; 5,000+ articles

      • GraphQL
      • MongoDB
      • Express & Node.js
      • TypeScript
      • RSS Parsing
      GitHub
    • PriceMerge screenshot

      PriceMerge

      PriceMerge empowers users to combat price hiking by allowing users to compare and track product prices across thousands of websites, automatically notifying them when prices drop. PriceMerge aided tens of thousands of users in acquiring PPE during the COVID-19 pandemic and has been featured on the national news.

      25,000+ Users; 200,000+ Searches; Featured on national news

      GitHub
    • Eatery: Cornell Dining screenshot

      Eatery: Cornell Dining

      Eatery is an app built by Cornell AppDev to display Cornell dining hall menus to 10,000+ users daily. I worked as a backend developer on a team of 10 to architect a new backend from scratch to enable account personalization, food recommendations, and item favoriting.

      10,000 Monthly Users; 30,000+ downloads

      • Django
      • PostGreSQL
      • SQLAlchemy
      GitHub

    Areas of expertise.

    Across the stack, from model training to application building, these are the tools I’ve leveraged to build production ready solutions.

    • Next.js
    • React.js
    • Svelte
    • Tailwind CSS
    • Node.js
    • Express.js
    • Flask
    • Django
    • MongoDB
    • GraphQL
    • PostgreSQL
    • MySQL
    • Docker
    • Kubernetes
    • Terraform
    • AWS, Azure, GCP
    • PyTorch
    • Hugging Face
    • TensorFlow
    • SKLearn
    • React Native + Expo
    • Swift
    • Objective C
    • Jetpack Compose
    • Java
    • Python
    • C++
    • JavaScript & TypeScript

    • Frontend

      • Next.js

        Rendering Framework

      • React.js

        Web UI Library

      • Svelte

        Web UI Library

      • Tailwind CSS

        CSS Framework

    • Backend

      • Node.js

        JavaScript runtime

      • Express.js

        JS Backend Framework

      • Flask

        Backend Framework

      • Django

        Backend Framework

    • Databases

      • MongoDB

        NoSQL Database

      • GraphQL

        API Query Language

      • PostgreSQL

        Relational Database

      • MySQL

        Relational Database

    • DevOps

      • Docker

        Container Creation

      • Kubernetes

        Container Orchestration

      • Terraform

        Infra-as-a-Service

      • AWS, Azure, GCP

        Cloud Services

    • Machine Learning

      • PyTorch

        Deep Learning Framework

      • Hugging Face

        Transformer Library

      • TensorFlow

        Deep Learning Framework

      • SKLearn

        ML Toolkit

    • Mobile

      • React Native + Expo

        Mobile Dev Framework

      • Swift

        iOS Development

      • Objective C

        iOS & MacOS Development

      • Jetpack Compose

        Android Development

    • Languages

      • Java

        Highly Experienced

      • Python

        Highly Experienced

      • C++

        Experienced

      • JavaScript & TypeScript

        Experienced

    After hours

    The Study: A Peek Into My Brain.

    Check out the Study to learn about the best books I’ve read, the research questions I’m currently pondeirng, and notes from the academic side of my life.

    Is there a fundamental limit to the generalization capability of Transformer based Large Language Models?
    Visit the Study

    Browse the shelf

    Say hello.

    Building something ambitious or want to talk about what I'm up to? My inbox is open.

    Based in San Francisco, CA. Replies usually within a few days.