Adiyan Kaul

Developer
Adiyan Kaul

About me

Hi, I'm Adiyan Kaul, a full stack developer and entrepreneur. I specialize in AI research, web development, and product building. Over the years, I've developed expertise in a range of technologies including React, Python, JavaScript, and Rust. I'm passionate about solving complex problems through innovative technology solutions. As a Regents Scholar from UC Berkeley and an Eagle Scout, I bring both technical expertise and leadership experience to every project. When I'm not coding or building products, I enjoy playing basketball and reading books to better help me understand the world around me.

Experience

July 2024 - Present

Cofounder @ Dubs

San Francisco, CA

Building conversational AI toys with personalized interactive experiences. My team and I developed the entire technology stack from hardware to software, including 3D printing and injection molding the physical toy with an integrated ESP32 chip, speaker, and microphone. The backend consists of a custom Rust server with a sophisticated AI pipeline featuring voice activity detection, speech-to-text processing, a self-hosted fine-tuned large language model, memory system for RAG on the LLM and text-to-speech synthesis. I also built ESP32 firmware handling Bluetooth and WiFi connectivity with acoustic echo cancellation, plus an Expo React Native companion app for WiFi setup, battery monitoring, and device control. After successfully pitching at Toy Previews LA, we're now in negotiations with multiple toy companies for licensing deals to integrate our conversational AI technology into new product lines.

June 2021 - July 2024

Full Stack Software Engineer 3 @ FloQast

Los Angeles, CA

As an engineer at FloQast, I built and maintained React frontends and AWS Lambda backend APIs that controlled user access to critical accounting workflows including close management, operations, and payroll systems. These systems served over 3,000 enterprise clients including major Fortune 500 companies. I led several large-scale initiatives that directly impacted the company's growth strategy, including developing an ad-hoc project management workflow system that improved product stickiness as FloQast expanded beyond its core close management offering. I also spearheaded a complete navigation system overhaul to support the company's rapid product expansion. Throughout my tenure I also did work in CI/CD, QA testing, mentoring new hires, product development, customer support, and collaborating with cross-functional teams to ensure seamless integration of new features.

May 2020 - Aug 2020

Cloud Engineering Intern @ McAfee

Santa Clara, CA

During my internship with McAfee's cloud engineering team, I conducted a comprehensive analysis to determine the optimal database solution between Aurora and DynamoDB for our specific data requirements. I created detailed database schemas, populated them with realistic test data, and ran extensive sample queries to perform a thorough cost-benefit analysis. Additionally, I built a real-time notification web application using Firebase, Aurora, Python, Amazon SNS, and AWS Lambda that could instantly push notifications to all of a user's devices whenever changes occurred in the database. This system improved user engagement and provided immediate feedback for critical security events.

May 2019 - Aug 2019

Data Science Intern @ McAfee

Santa Clara, CA

As a data science intern, I developed a sophisticated machine learning solution to combat Android malware. I implemented a random forest algorithm in Python that could determine if an Android app was malware with 90% accuracy by analyzing features such as the permissions requested by the app, developer information, and the trust factor of the app's network endpoints. I also worked on internet traffic anomaly detection for IoT devices using Apache Spark to model network traffic trends and identify patterns that could indicate attacks such as denial of service attempts. This work contributed to McAfee's mobile security offerings and helped protect millions of users from malicious applications.

June 2017 - Aug 2017

Intern @ NASA Ames Research Center

Mountain View, CA

As an intern at NASA I worked on a project involving autonomous drones and machine learning. I wrote a convolutional neural network able to distinguish people with suspicious attributes, like concealed weapons and backpacks. I also worked on the hardware aspect of mounting objects on the drone like a camera and a raspberry pi. The software was then loaded on the drone so that it could look for and track individuals using the algorithm. This was presented to a congressman, the Army, and several senior individuals at NASA.

June 2016 - Aug 2016

Software Intern @ Keysight Technologies

Cupertino, CA

As an intern in the Keysight Labs division I worked on evaluating modulation recognition using machine learning algorithms. The tasks I had to accomplish included acquiring over-the-air RF signals, developing software to evaluate the accuracy of different models, running tests, and reporting test results.I was able to obtain three different types of over the air signals and train a neural network to be able to distinguish between different signal spectrograms. I also developed a web service that enable non expert users to upload and classify signals and a service to train models and hosted it on an Amazon EC2 instance.

Education

Aug 2018 - May 2021

B.S. from EECS @ UC Berkeley

Regents Scholar (Top 200 Incoming Freshmen)

Graduated in 2021 with a Bachelor's degree in EECS. I took courses in electrical engineering, computer science, math, and history. These courses helped me gain a deeper understanding of machine learning, computer architecture, algorithms, data structures, databases, and systems design.

Aug 2014 - June 2018

Diploma @ Cupertino High School

Cupertino, California

I graduated highschool where I enjoyed courses in math, computer science, and history. Throughout highschool I was always set on computer science and as a result I frequently took part in hackathons and science fairs. Won many awards in these competitions. I also participated in other extracurricular activities such as parlimentary debate and was ranked top 10 in the USA for this debate format as a junior.

Skills

Here is a summary of my most important skills and abilities as a Full Stack Developer:

Technical Skills

Python
Javascript / TypeScript
React Native
Rust
HTML / CSS
Node.js
Java
Swift
GraphQL
R
PostgreSQL
JSON / XML
Express.js
SQL
Angular JS
C++
CI / CD

Languages

Kashmiri (mother tongue)
English (daily use)
Hindi
Spanish

Other

AWS
Github
Figma
Jira
Scrum / Agile
Docker
SEO
Postman
CLI
MacOS / Linux
Grafana

Projects

My favorite projects I've worked on in my career that are worth sharing.

Risq

Risq - Cofounder

Risq was a competitive paper trading platform where users could compete against each other in stock market simulations with real payouts based on their performance and the amount of money they paid to play. The app gamified investing education by creating tournaments and leaderboards that motivated users to learn about financial markets while competing for prizes. My partner and I built the entire technical infrastructure using React for the frontend and AWS Amplify for the backend, integrating GraphQL APIs, DynamoDB for data storage, Lambda functions for serverless computing, and IEX Cloud for real-time market data. We successfully pitched to venture capitalists through Stanford ASES and secured multiple follow-up meetings, validating our product-market fit.

Link Offline
Technologies React, AWS Amplify, GraphQL, DynamoDB, AWS Lambda, AWS API Gateway, IEX Cloud API
Hawkeye

Hawkeye - California State Science Fair

My partner and I worked on a project which used drones to assist autonomously with search and rescue missions. We mounted a Raspberry Pi and a Raspberry pi camera on a 3DR Y6 drone, wrote code to identify people using computer vision, and implemented the D*Lite pathfinding algorithm in C++ and Python to guide a drone in an unfamiliar and dynamic environment to find the shortest path possible to reach the safe locations to help navigate people to safety. We also developed an interface which would allow multiple drones to work together to quickly and efficiently to complete the mission to save lives. This project won 1st award in the Physical Science and Engineering Category at the regional Synopsis Science Fair in Santa Clara, California and 4th place in the Engineering category in the California State Science Fair.

Link Paper
Technologies Python, C++, Dronekit, OpenCV
Dermyx

Dermyx - Health++ Stanford Hackathon

Dermyx was a mobile app built with Electron and Swift that rapidly detected melanoma skin cancer. From the comfort of one's home, an individual could take a quick picture of an area on his or her skin inside the Dermyx app. Then the algorithm (which used a machine learning algorithm trained on a database of malignant and benign melanoma pictures) operating in the background of Dermyx would discern and classify the picture taken as either benign or malignant. I also implemented a RESTful API that allowed anyone to send a picture of their skin and get a diagnosis without the use of the app. We won the Best Global Oncology Prize at this hackathon for this project and were the only high schoolers to win a prize at the hackathon as the hackathon was primarily for health care professionals and PHD students!

Technologies Python, Flask, Electron, Swift, Node.js, IBM Watson, JavaScript
Vitruvian

Vitruvian - HSHacks 3

Vitruvian was a simple and elegant mobile application designed to help doctors in developing areas keep track of patients and diagnose diseases. Inspired by the lack of centralized medical data, it aimed to simplify healthcare delivery and make critical information more accessible. The app featured centralized patient records, a machine learning model that I wrote which predicted heart disease risk with 72% accuracy, and an SMS-based assistant that provided medical guidance to users without internet access. Despite challenges with data storage and optimization, Vitruvian demonstrated the potential of accessible, data-driven healthcare tools in resource-limited settings. We won multiple awards at the hackathon for this innovative solution.

Technologies Python, Swift, Node.js, Flask, Keras, Twilio, Redis, Alamofire, NLP, Convolutional Neural Networks
swift

SwiftAssist - Los Altos Hacks

SwiftAssist crowdsourced emergency responses so that people who carry allergy medication or EpiPens, people who are CPR certified, or doctors can immediately respond to emergencies in their vicinity, faster than 911. My team and I developed an iphone app, a web app, and a pebble app so users could register, request help, and respond to help requests all with a single click. People in dire need of help could send for help through a Pebble watch or the iOS app with an easily accessible button which would then notify every person in the area who is certified to help and can reach the location faster than 911 to help save lives. We won an award for Best Use of Microsoft Technologies at the hackathon with overwhelming positive feedback on the platform.

Technologies Swift, Electron, JavaScript, Pebble.js, Azure, Firebase, Photon