About

I’m Nicole Hao, a MEng student in Computer Science at Cornell University. Previously, I received my B.A. in Mathematics from Cornell in Dec 2024.

Professionally, I work at the intersection of R&D and GTM strategy, using multimodal AI and RAG to help people improve their productivity and help enterprises scale.

I come from an academic research background in mathematics, machine learning, and astrophysics, with a focus on building applied AI systems that bridge theory and real-world use cases. My work integrates multimodal data, including language, vision, and unstructured textual information, to power scalable AI solutions.

Outside of work, I lead several nonprofit initiatives focused on advancing accessibility in education.

“Education is not a priviledge, it is a human right.”

As a first-generation college student, I’ve witnessed countless brilliant young minds being excluded from educational opportunities throughout my life, not because of a lack of potential, but because of systemic limitations.

This is why I am an ardent advocate for educational access through EdTech. I focus on designing, building, and advertising tools that integrate VLMs, RAG, Robotics, and, most importantly, human compassion, to make a difference in education.

I believe that with the help of technology, we will one day augment human cognition and help build a more equitable world, one where physical or cognitive challenges no longer limit a person’s potential.

Fun fact: I go on mediation retreats. No, don’t get me started on Buddhism (and my screen time).

Research Projects

I’ve led both engineering and research projects, grounded in the belief that impactful work begins with innovation, and research is at the heart of that.

My research centers on applying math and AI/ML to solve real-world problems in science and education.

  • InkSight AI
    Do you know that many students with visual impairments are charged as much as $100 per hour for note-taking assistance, just to access the same classroom content as everyone else?

    This is why me, Arjun Maitra (Cornell Math & Stats 28’), and Clément Rozé (Cornell Info Sci 28’) did months of research and developed InkSight AI. InkSight AI is an education platform that captures information across modalities, making real-time, multi-sensory learning possible and affordable for every learner. I was lucky enough to be selected by the Cornell eLab Student Startup Accelerator as part of their Fall 24 cohort with this project.
    Work in Progress. Demo coming soon.

  • Multimodal STEM Lecture Video Dataset & Data Labeling Tool
    I developed a multimodal STEM lecture video dataset and a custom data labeling tool to support models ranging from FCNs to VLMs that interpret complex classroom content. The tool enables precise annotation of visual, textual, and auditory elements, such as handwritten equations, diagrams, and spoken explanations, laying the groundwork for accessible, multisensory learning systems. I was advised by Prof. Jennifer Sun for this project.
    Work in Progress. Demo & Paper coming soon.

  • Operator Learning in Sobolev Spaces
    This project bridges functional analysis and operator learning to (more rigorously) analyze how deep neural networks approximate nonlinear operators between Sobolev spaces. I built on the universal approximation in Sobolev norms and derived an original theorem that gives an explicit bound on the neural network complexity to approximate a nonlinear operator between Sobolev spaces.

  • Evaluations on Response-Based Knowledge Distillation and the Effects of Alpha on Model Accuracy
    I conducted this research project under the guidance of Prof. Yunan Yang to investigate the effects of the alpha parameter in response-based knowledge distillation using MNIST data. By implementing a custom distillation framework in Keras, I analyzed how varying alpha values influence student model accuracy in contrast to direct learning from data and guidance from a teacher model. The study provides more information on optimizing model compression and generalization performance in low-resource settings.

  • Classifying Solar Flares Using Supervised ML
    As a Nexus Research scholar at Cornell, I developed a novel supervised learning pipeline to detect and classify solar flares using high-resolution solar spectra from the RHESSI and HARPS-N missions under the mentorship of Prof. Ray Jayawardhana and Dr. Laura Flagg. We trained several machine learning models and found that a Support Vector Classifier with an RBF kernel achieved the highest performance. Our work shows that even subtle spectral changes from weak flares can be detected algorithmically, paving the way for future applications in mitigating stellar contamination in exoplanet studies.

    Our research results were published in the Astrophysical Journal.

Engineering Projects

To me, innovation is not just about research - building a viable product and gaining traction are equally important.

My engineering projects focus on business and finance, specifically, using AI to improve productivity and workflows.

  • LeadGen.AI
    An AI-powered sales search & outreach platform for targeted B2B leads based on your ideal customer profile. The system automates finding and qualifying sales leads by researching industry events and trade associations where potential customers might be present. ⭐ Check out the new Demo and loom walkthrough

  • Cornell AI Innovation Lab Projects
    I was selected to work at the Cornell AI Innovation lab. I was involved in different projects building RAG-based AI agents that improve productivity in the expenses review process, and offer AI-generated second-opinions at the Cornell Agricultural and Life Sciences School and the Cornell Vet School.

  • AI Engineering at Ascot Group
    At Ascot Group, an international specialty reinsurance company, I developed AI-driven tools to improve decision-making for underwriters. I built a machine learning system that scores broker submissions on a 1–5 scale based on profitability and win-rate indicators, and fine-tuned NLP models to extract structured data from unstandardized PDF loss runs. Through this work, I witnessed firsthand the inefficiencies that arise from complex, manual processes in the financial industry. This experience deepened my interest in building AI systems that simplify workflows and reduce operational overhead.

Talks & Presentations

I LOVE pitching and giving speeches! Nothing excites me more than getting on stage in front of a crowd and presenting a novel idea, product, or research findings.

Volunteering & Community Service

One of life’s most crucial aspects is giving back to the communities that helped you.

It hurts me to see younger people going through the same struggles I once had to go through, so I try to help people pursue education opportunities and find a sense of belonging through community service.

  • Volunteer, Dharma Drum Retreat Center, May - Jul 2025

  • Resident Advisor, Holland International Living Center, Aug 2022 - Dec 2024

  • Volunteer Mathematics Tutor, GoPeer, Aug 2021 - Dec 2023

  • More to be added: I have been trying to use my public speaking skills more for advocacy projects. If my skillset fits your goals, reach out to me!

Recent Reads

  • The Second Half - A great blog by Shunyu Yao on the development of AI, and the shift to evaluation setup in terms of the most important problem of AI.
  • How to Do Great Work - An amazing essay by Paul Graham. When in doubt, “stay upwind”. This phrase stuck with me.
  • The Meaning of Meaningless: The Importance of Meaninglessness - By Dr. Nanige Nashiko. I don’t think there is an English translation for this book yet. It’s a great book about being able to derive joy from doing meaningless activities in a meaning-driven world. 「無意味を楽しむことが、最も意味のある生き方だ。」Finding joy in the meaningless is the most meaningful way to live.
  • Agency is Eating the World - by Gian Segato. “AI has eroded the value of specialization because, for many tasks, achieving the outcome of several years of experience now takes a $20 ChatGPT subscription.” Agency has become one of the most desirable traits among working professionals.
  • More to be added…

Last updated: May 2025