Gary Gao ('28), a sophomore in the College and in Wharton, spent this past summer in Professor Marc Miskin's Circuits Lab. There, Gary helped design and prototype a oscillator-based physical learning network, a circuit that can perform AI tasks for a fraction of the energy. This internship gave Gary the opportunity to learn the processes necessary for neuromorphic computing. This experience was supported by the Franklin Opportunity Fund, an internship funding opportunity for Benjamin Franklin Scholars.
Gary Gao – SAS ’28, W’28
This summer, I worked in Professor Marc Miskin’s circuits lab to design and prototype a novel oscillator-based physical learning network – a circuit that can perform AI tasks like regression or classification while using a fraction of the speed and energy of a digital counterpart.
I worked closely with Dr. Marc Miskin, Adam Frim, and other members of the lab, and my role spanned three concrete pieces: formalizing mathematical learning rules, prototyping breadboard circuitry, and presenting results during group meetings.
I didn’t “find” this internship through a conventional search – I’d already been researching in the lab for five months, and I was given the opportunity to continue over the summer.
Coming in, I was excited because the lab had already let me build real momentum. Over the prior semesters I had the chance to research network topologies, program Python interfaces, and help build an end-to-end cloud-computing platform for physical learning, while also developing practical skills like rapid prototyping.
The bigger “why” for me is the promise and peril of AI. As a researcher and developer who is applying AI in various contexts, such as pottery classification or data collection, I understand the benefits that AI can have when used correctly. However, as an ISP student, I also recognize the environmental and social costs that energy-hungry datacenters can have. Developing the next generation of energy-efficient AI infrastructure allows me to bridge this gap and take part in creating a sustainable AI future where we can harness its benefits without absorbing unnecessary harms.
Pictured: A Nine-Node Oscillator Network
Day-to-day, the work followed a disciplined pipeline. We started with background research and formalizing the learning rules, and the first step in these circuit projects was always to validate those rules by simulating them in Jupyter Notebooks. Only after confirming the simulations did we move into hardware – building the oscillators and eventually the learning edges.
Not everything went smoothly, and that was the point. Early on, I tried to move too quickly in the first two weeks and missed some critical waveform tests, which created significant oscillator errors. Catching the issue quickly still taught me the lesson I needed – to verify each step to a high standard before moving on.
I also learned how time-intensive hands-on research can be. Waiting days for parts to arrive, especially when building learning edges that required multiple integrated-circuit components, made it hard to plan for time efficiency while simultaneously building and discovering.
What kept me going was the joy of seeing the circuits learn, and the lab community. Hearing other researchers’ progress every Thursday inspired me and broadened my view of neuromorphic computing.
Going forward, I plan to carry the patience and rigorous testing mindset I gained this summer as I keep developing these physical learning networks, in service of a more sustainable AI future.
Interested in reading more accounts from BFS? Click Here
Related Articles
Using Data Science Address Challenges and to Improve Patient outcomes: Interning at the Children's Hospital of Philadelphia
Evan Grove ('26), a System Science & Engineering major, spent this past summer serving as a Data Science Intern at the Children's Hospital of Philadelphia, working in the Translational Imaging Research Unit in the Center for Data Driven Discovery in Biomedicine. In this internship Evan worked to develop and apply artificial intelligence to better classify pediatric brain tumors. This experience was supported by the Franklin Opportunity Fund, an internship funding opportunity for Benjamin Franklin Scholars.
Legal and Policy Research in Practice: A Summer at the University of Minho
Dalila Gafar ('26), a double major in Legal Studies & Business Ethics and Operations, Information, & Decisions, spent this past summer at the University of Minho’s R&D Unit, the Center for Justice and Governance. The internship gave Dalila the opportunity to engage in legal research and compose a report on the EU Digital Services Act. This experience was supported by the Franklin Opportunity Fund, an internship funding opportunity for Benjamin Franklin Scholars.