Sunjung Lee

I completed the 2026 CureSearch Ultimate Hike on September 19th, hiking 21 miles on the beautiful Colorado Trail from the South Platte River to Meadows Group Campground. Hikers, coaches, and volunteers came together to raise nearly $208,000 for pediatric cancer research. Watching my fellow hikers complete those 21 miles, with volunteers and coaches supporting them every step of the way, was nothing short of inspiring. Listening to the stories of the children who have fought through this dreaded disease—and honoring those who tragically could not—was both humbling and motivating. Thanks to the generous donations from my family and friends, I have been able to raise $12,172 so far. But the need is real, and it is urgent. So much more needs to be done so that no child ever has to face this reality. You can still donate using the link below. Thank you for helping me tell children's cancer to take a hike! CureSearch for Children's Cancer Lauren Rump Alyssa Cooling, CNP https://lnkd.in/gv4XF7ix

Just attended the AI4 2026 conference last week — keynotes, workshops, expo floor, and countless conversations with builders across the AI stack, from infrastructure to data platforms to AI applications. A few takeaways I keep coming back to: 1. AI is a tool — a uniquely powerful one. The strongest teams aren't applying AI for the sake of it. They keep the deterministic flows that work intact, identify the real gaps, and focus AI on the problems truly worth solving. As Andrew Ng noted, let AI go back into the background like electricity, while we humans use it effectively and perform our tasks. 2. The 50x delta is real. One point from the keynotes stuck with me: the performance gap between those who master AI and those who don't won't be 10% or 20% — it will be 50x. Some tasks will be replaced, virtually all jobs will be reconfigured, and many more will be created by the sheer amount of new opportunities AI enables. Education — not just curriculum, but sustaining the motivation and urgency to learn — will matter more than ever. 3. AI is becoming part of the engineering process itself. Just as we formed DevOps to embed CI/CD into how we build and release products, AI will be integrated into the entire engineering lifecycle — requirements, design, development, and deployment. One example that stood out: a physics foundation model for hardware design — an intelligent layer that understands thermal behavior, layout, and board configuration. Imagine how many other world models are waiting to be built. 4. Own the data — and the end-to-end stack. As an AI solutions engineer working on the end-to-end development of GenAI applications, this theme resonated deeply. The expo floor was full of startups serving every layer — infrastructure, model management, AI services, orchestration, deployment pipelines. But for a technology company, the strategic advantage comes from owning the entire stack: guardrails defined per use case, customizations for cost and performance, and the flexibility to reroute models as needs evolve. Build vs. buy isn't just about deployment speed — it's about risk management and flexibility. One panelist shared how a SOTA vendor solution from two years ago, and enabled the team with faster speed at the time, but became something their team could build in-house & cheaper — and they learned the stack in the process. 5. Production agents are an engineering discipline. The most valuable sessions weren't about raw model capability, but the craft around it: designing context and memory deliberately, deciding how much intelligence a task is actually worth, setting guardrails, and using agent traces to understand why a decision was made. And on measuring success — one honest outcome metric beats ten vanity ones. Working on Edge AI and agentic workflows every day, I left energized: the frontier isn't just bigger models — it's the discipline of putting intelligence to work where it truly creates value. #AI4 #AI #EdgeAI #Qualcomm

Excited to share our recent work published in Nature Communications! “A permeable protein nanocage enables facile cargo loading and cytosolic protein delivery” Paper link: https://lnkd.in/gwcJqHzC In this study, we discovered that the encapsulin from Quasibacillus thermotolerans (QtEnc) possesses unprecedented intrinsic permeability, enabling efficient post-assembly internalization of diverse protein cargos in vitro. Leveraging this property, we engineered a QtEnc-based NanoCarrier (QtEncNC) that achieves efficient cytosolic protein delivery without requiring shell disassembly, establishing a promising platform for intracellular therapeutic delivery. Huge thanks to my advisor, Prof. Tobias Giessen, and to Mike Andreas and Jesse Jones for their invaluable contributions and support throughout this project. Alongside this exciting milestone, I’m also thrilled to share that I’ll be joining the Institute for Protein Design, University of Washington, as a postdoctoral researcher!

Excited to share that our paper has been accepted at the 2026 IEEE Symposium on VLSI Technology and Circuits — and honored that it was nominated for the VLSI Best Student Paper Award! Smart textiles are evolving beyond passive garments into active systems that can sense, process, and respond to the environment in real time. The application space is broad, from healthcare to industrial safety, but a persistent challenge remains: How do we make these systems small and long-lasting? We present a fully integrated video encoder system with a custom 2x2 microlens array, designed for small, energy-constrained wearables. With a compression algorithm and hardware tailored for the 2x2 microlens array, the system achieves: - A compact 31.5 × 10.7 × 4 mm³ form factor - A 90.6 pJ/pixel minimum ISP energy at 1280×800, 65fps - Longest system recording time with system Flash storage among state-of-the-art works - Best compression-image quality tradeoff among the state-of-the-art works I’m grateful to my collaborators and advisors for their support throughout this project. Looking forward to presenting the work at the VLSI Symposium in Hawaii next week! https://lnkd.in/gY3gM2fq #VLSI2026 #WearableTech #ASIC

This September, I’m hiking 20+ miles in one day to raise critical funds for children facing cancer. All proceeds support CureSearch for Children’s Cancer and their work to fund safer treatments and faster cures for kids and families who need them most. This is my 6th year supporting CureSearch, and I’ve already raised nearly $10,000 this year. If you’re able, please consider donating or sharing this with your network, every bit of support helps move us closer to better outcomes for children with cancer. Thank you for helping me tell children's cancer to take a hike! https://lnkd.in/gv4XF7ix

As 2025 comes to a close, we’re reflecting on what you made possible this year. Thanks to your generosity, we fueled groundbreaking research, supported families facing childhood cancer, and brought promising treatments closer to the children who need them most. Every gift, every step, every voice mattered, and we couldn’t have done it without you. Here’s to continuing the momentum together in 2026! 💚

🎓Congratulations to Jungho Lee on a successful PhD defense! Jungho’s research explores the frontiers of ultra-low-power microsystems, tackling the circuit and system challenges needed for tiny bioimplants and microrobots. His work includes wireless neural recording and stimulation chips that use just a few microwatts of power, as well as an optically reprogrammable microrobot that combines energy harvesting, sensing, computation, and autonomous behavior. Together, these achievements help push the Internet of Things beyond traditional devices, opening new possibilities for biointerfaces, medical implants, and intelligent microrobotic systems. Jungho was advised by Prof. David Blaauw University of Michigan College of Engineering