Welcome, my name is Anna!

This is a space where I share my authentic self—my achievements, fears, failures, aspirations, thoughts, and everything in between. My goal is that as you explore this page, you'll not only get to know me better but also find something that inspires you to reach out via DM or email to connect.

Welcome, my name is Anna!

When it comes to my day job I love all things data and machine learning, but on the weekends I enjoy exploring HCI and the more creative side of technology. I have a solid foundation in machine learning and data engineering from the University of Michigan, I have navigated roles as diverse as Design Engineer, Trading Analyst, and XR Dev and Lab Manager. My research work spans from analyzing carbon emissions in underdeveloped countries to studying the impact of deep learning in computational finance and exoplanet characterization.

Throughout 2023, I travelled across the country every two weeks to participate in hackathons. These hackathons were not just competitions; they were my classroom, keeping me abreast of cutting-edge tools, technologies, and opportunities in tech.

I've set my sights on upskilling and earning better credentials in 2024, aiming to enhance my credibility and create a solid foundation for turning ambitious plans into successful outcomes. Whether you're a fellow tech enthusiast, potential collaborator, or just passing through, I'm glad you're here. Let's connect and explore how we can make a difference, together.

Current Major Projects

Gralx - multi-modal data mask that preserves context

Gralx is an open-source project that preserves vital contextual information in multi-modal training datasets while anonymizing personally identifiable information (PII) through pseudonymization with synthetic data. Traditional data masking often removes crucial context—like race, gender, age, and emotions—essential for training accurate and unbiased AI models. Gralx allows users to upload multimedia data and specify which contexts to retain. It then generates synthetic data that maintains these important attributes but replaces sensitive identifiers. By integrating advanced models for dense video captioning, audio transcription, text-to-image generation, and image-to-video animation, Gralx creates a pipeline that produces PII-compliant datasets ready for machine learning without sacrificing essential context.

Drolg - AI party DJ by processing video feed in real-time

Drolg is a project that helps DJs choose the perfect music for their audience by analyzing crowd movements and reactions using computer vision. By processing live video feeds, the system interprets the crowd's energy and vibe—assessing factors like movement and pose—and selects songs from Spotify that match these dynamics. This ensures that the music resonates with the audience's mood in real time, enhancing their overall experience and keeping the energy high on the dance floor.

• Jane Street Real-Time Market Data Forecasting | 1/6/25
• Google Unlock Global Communication with Gemma | 1/14/25
• AI Mathematical Olympiad - Progress Prize 2 | 4/1/25

Professional Goals

• Google
• NeurIPS, attend, paper accepted
• Masters
• Media Lab @ MIT
• Quant
• Manufacturing
• Fluent Chinese
• Archeologist

Personal Health Goals

• Lose 40 lbs
• Lose 120 lbs
• Lose 120 lbs
• Run 5K

Adventure & Travel Goals

• Hike Volcano
• Skydiving License
• Bass Canyon
• Burning Man
• Turkey
• Canyonlands

Creative & Artistic Goals

• Make Art Creative Technology Code
• Artist Collective

Financial Goals

• Invest in Emerging Markets
• Sell Company and "Retire"

Social Goals

• Social Experiments
• Social Club

Lifestyle Goals

• My Dream Farm