Vellex Computing

For International Women’s Day, we asked the women founders building hard tech ventures in our fellowship how they lead. What they described wasn't a list of traits. It was a blueprint. 🔹 Communication as infrastructure 🔹 Values as system requirements 🔹 Deliberation as technical precision Etosha (Eee-tah-sha) Cave (Twelve, Cohort 2015) designs communication with the same rigor as the science itself. Ilse Nava-Medina (Gel Matter, Cohort 2023) and Palak Jain (Vellex Computing, Cohort 2023) build values into the architecture from day one. Sarah Placella (Root Applied Sciences, Cohort 2022) brings the precision of molecular biology to every product decision. This isn't leadership adapted to fit a system. It's systems thinking applied to company-building. Values-driven hard tech isn’t about choosing mission over returns. It’s about reducing risk early, building stronger teams, and increasing the odds that complex science actually makes it to market. We’re proud to support founders who are building companies designed to last—and deliver. Read the full piece: https://lnkd.in/gRCbQYWg

For the past couple of years, I've been attending conferences focused on one urgent question: 𝗵𝗼𝘄 𝗱𝗼 𝘄𝗲 𝗽𝗼𝘄𝗲𝗿 𝘁𝗵𝗲 𝗔𝗜 𝗿𝗲𝘃𝗼𝗹𝘂𝘁𝗶𝗼𝗻? The story is familiar by now — 𝗱𝗮𝘁𝗮 𝗰𝗲𝗻𝘁𝗲𝗿𝘀 𝗮𝗿𝗲 𝗵𝗶𝘁𝘁𝗶𝗻𝗴 𝗽𝗼𝘄𝗲𝗿 𝘄𝗮𝗹𝗹𝘀, the electric grid is facing massive backlogs, and 𝗲𝗻𝗲𝗿𝗴𝘆 𝗵𝗮𝘀 𝗯𝗲𝗰𝗼𝗺𝗲 𝘁𝗵𝗲 𝘀𝗶𝗻𝗴𝗹𝗲 𝗯𝗶𝗴𝗴𝗲𝘀𝘁 𝗯𝗼𝘁𝘁𝗹𝗲𝗻𝗲𝗰𝗸 𝗳𝗼𝗿 𝘀𝗰𝗮𝗹𝗶𝗻𝗴 𝗔𝗜. The energy industry is racing to ramp up power plants and build out new infrastructure just to keep up. But this year, attending 𝗜𝗦𝗦𝗖𝗖 𝟮𝟬𝟮𝟲 gave me a completely different perspective — one from the other side of the equation. AI models have exploded in size, but compute, memory, and interconnect haven't kept pace. CPUs and GPUs are general-purpose by design and largely underutilized for the specialized workloads AI demands. So while the energy industry is trying to supply more power, the semiconductor community has been asking a different question: 𝘄𝗵𝗮𝘁 𝗶𝗳 𝘄𝗲 𝗷𝘂𝘀𝘁 𝗻𝗲𝗲𝗱 𝗳𝗮𝗿 𝗹𝗲𝘀𝘀 𝗼𝗳 𝗶𝘁? And they're delivering. Here are the key shifts I saw at ISSCC: • 𝗦𝗽𝗲𝗰𝗶𝗮𝗹𝗶𝘇𝗲𝗱 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲𝘀 𝗮𝗿𝗲 𝗿𝗲𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝘁𝗵𝗲 𝗿𝘂𝗹𝗲𝘀: From analog in-memory computing chips to task-specific accelerators, this community has spun off a wave of chips showing tremendous improvements in power consumption, efficiency, and raw compute speed. • 𝗦𝗰𝗮𝗹𝗶𝗻𝗴 𝘂𝗽 𝗮𝗻𝗱 𝘀𝗰𝗮𝗹𝗶𝗻𝗴 𝗼𝘂𝘁: Hardware architectures are becoming increasingly disaggregated, with each IP block performing a different specialized task. Even for LLM inference, prefill is handled by one piece of hardware and decode by another. Chips are being 𝗰𝗼-𝗱𝗲𝘀𝗶𝗴𝗻𝗲𝗱 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗮𝗹𝗴𝗼𝗿𝗶𝘁𝗵𝗺𝘀 𝘁𝗵𝗲𝘆 𝗿𝘂𝗻, and hardware-aware training is becoming a key strategy to make AI’s energy demands manageable. • 𝗧𝗵𝗲 𝗽𝗼𝘄𝗲𝗿𝗳𝘂𝗹 𝘁𝘄𝗼-𝘄𝗮𝘆 𝘀𝘁𝗿𝗲𝗲𝘁: On one side, we’re advancing AI accelerators and specialized hardware through novel circuit design and analog compute. On the other, 𝗔𝗜 𝗶𝘁𝘀𝗲𝗹𝗳 𝗶𝘀 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗻𝗴 𝘁𝗵𝗲 𝗱𝗲𝘀𝗶𝗴𝗻, 𝘃𝗲𝗿𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻, 𝗮𝗻𝗱 𝗹𝗮𝘆𝗼𝘂𝘁 of those very circuits — including analog designs that have traditionally been among the hardest and most time-consuming to build. These two forces are feeding each other, and the pace is accelerating fast. I’m returning home more convinced than ever about what we’re building at Vellex Computing. Our mission is to enable 𝗔𝗜 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗼𝗻 𝗺𝗶𝗹𝗹𝗶𝘄𝗮𝘁𝘁𝘀 𝗼𝗳 𝗽𝗼𝘄𝗲𝗿 — true on-device adaptation and intelligence. We eliminate the slow, power-hungry cloud round trips needed to update model weights for inference on edge devices. 𝗪𝗲 𝗯𝗿𝗶𝗻𝗴 𝗚𝗣𝗨-𝗹𝗲𝘃𝗲𝗹 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗼𝗻 𝘁𝗵𝗲 𝗲𝗻𝗲𝗿𝗴𝘆 𝗯𝘂𝗱𝗴𝗲𝘁 𝗼𝗳 𝗮 𝗰𝗼𝗶𝗻 𝗰𝗲𝗹𝗹 𝗯𝗮𝘁𝘁𝗲𝗿𝘆. #ISSCC2026 #AnalogCompute #AIHardware #SemiconductorInnovation #EdgeAI #Vellex

Stepping into a new chapter. I’m excited to share that I’ve joined Vellex Computing as a Technical Program Manager I. Vellex Computing is operating at the frontier of Analog Intelligence — transforming how real-world signals become faster, smarter AI decisions. The problems are challenging, the standards are high, and the opportunity to build meaningful technology is huge. Grateful for the trust, the warm welcome, and the smooth onboarding experience from the entire team. I’m looking forward to learning fast, taking ownership, and contributing to something truly impactful.