Parham Hajzavar

Owned the requirements gathering, full-stack redesign, development, testing, launch, and production support of the New Logon and Account Access web requests on GovConnect.iowa.gov, adding multi-layered security checkpoints and integrating an identity verification API that prevented malicious activity and spam attacks. With the new streamlined implementation, customers saved 24 - 48 hours per request approval. For a website of this size, that resulted in many dollars saved.

Developed and launched the UX flow, frontend development, and data transformation and transfer APIs for New and Renewal license applications for external agencies including the Iowa Alcoholic Beverages Division, Iowa Lottery, and the Department of Inspections and Appeals. The new business process resulting from these license applications modernized legacy PDF/paper forms and reduced total license processing time by 2 business days.

Throughout the one year project period, I worked closely with client program managers, business analysts, 3rd party vendor developers, training team, customer service department, and subject-matter experts to understand customer pain points and build the product requirements. These applications, used by 75% of Iowa businesses, resulted in a more streamlined and shortened licensing process for all stakeholders.

Worked at the focal point between subject-matter experts, internal agency staff, Legal, and program leadership to define the product requirements and build the tool that allows customers to receive real-time updates on their refund statuses, used by 750,000 taxpayers.

I teamed up with 3 MBA and a CS graduate student to compete in the UCLA Anderson School of Business Innovation Challenge under the Healthcare track. We won FIRST PLACE out of 18 teams with our invention of AnonNet, a synthetic data generation platform built using GANs and GenAI to artificially mimic the statistical characteristics of raw patient data, while achieving government-grade privacy measures. This lowers the barriers of entry and accelerates medical research.

I worked closely as a product manager alongside the core engineer behind the prototype, Shaan Mathur, and drove our team's business development initiatives by fine-tuning the product and our value proposition as per my research. I was the only undergraduate student in the competition.

• Interviewed over 40 medical researchers, venture capitalists, hospital executives, pharmaceutical executives, and business advisors to become the subject-matter expert for the business.
• Drove task delegation and research relating to the Business Plan, and personally authoring major sections like Product, Market, and Development & Operations on the deliverable Business Plan and our investor slide deck.
• Delivered the final pitch live to a panel of judges consisting of UCLA Anderson professors, venture capitalists, and healthcare tech executives.

GrubMatch solves the age-old problem of "Where should we eat?" by providing a gamified, Tinder-like swiping interface for restaurant selection that generates matches based on the preferences of all group members. This mobile app helps friends and couples quickly find restaurants they'll all enjoy!

Simply create a group, copy the 5-digit code to share with your swipe partners, and begin swiping asynchronously on restaurants within your input range. As soon as everyone in the group swipes right on the same restaurant, there's a Match, and the session comes to an end.

This app was built as a fun side project while in grad school, but I also cultivated my skills in product management, prompt engineering, UX/UI design, and native app development.

Stay tuned for the link! ⭐️

Note: No user data is collected or saved in the app, and does not generate any income – I only collect basic usage data to measure app performance.

App Screenshots

My team and I built an AI-powered education tool that allows instructors to upload recorded online Zoom lectures and receive an intuitive and actionable engagement analytics dashboard, helping pinpoint where students are most engaged (or lost). It is built using a custom-trained computer vision architecture (ResNet-50) that maps facial emotions to engagement levels.

The app is secure, self-contained, and does not expose our API. We do not store any student info, and after each session the lecture video that is uploaded by an instructor and any analysis generated using the video is wiped from our system, leaving us safe from most cyber attacks.

EngageAI System Overview