My name is Rob Ghilduta, I grew up in 🗽New York City but have been based out of 🌉San Francisco since 2012. I’m a tech founder, operator, researcher, and investor. I started in the information security industry in the mid 2000s, worked in tech (Meraki/Robinhood), founded tech companies, and currently invest.

Background:

I’m a textbook and arXiv geek. Since 2014 I have taken up reading an average of one textbook a month to build a foundation in linear algebra, applied mathematics, statistics, digital wireless communications, semiconductor design, and deep learning. I have architected agile narrowband and wideband RF modems. I fine tuned my first GPT model in 2019 for a scheduling assistant, and have been hooked on LLM research published on arXiv ever since. During the pandemic, I spent time studying molecular biology and immunology with the intent of speeding up molecular dynamics with learned optimizers. More recently I have been experimenting with agent-based models and embodied agents for a variety of research projects, and helping early stage companies.

In the corporate world, I worked on several high volume enterprise network devices, MEMS and compute intensive ASICs, developed resilient broadband communication systems, and helped optimize and secure large cloud distributed systems.

Practically, I’ve FEM’d PCBs and ASICs, CFD’d BLDC EDF motor RC airplanes, FEA’d CNC’d aluminum parts, improved STA of RTL for OFDM waveforms, SVD’d preambles for MIMO channel estimates, GROMACS’d Kinesin PDBs, rooted MCUs through XHCIs, WinDbg’d KMDF driver bugs over UART, SMT’d Sudoku puzzles, and fine-tuned GPTs on my IMAP.

Conversational and/or literate in several languages including 🇷🇴 Romanian, 🇩🇪 German, 🇨🇳 Mandarin (中文). At one point spoke 🇭🇺 Hungarian, 🇸🇰 Slovakian, and studied 🇫🇷 French, 🇪🇸 Spanish, and Latin. I’ll likely try to learn more languages to test new learning strategies.

Current:

• Maintain the bladeRF Software Defined Radio project, and run Nuand LLC
• Strategic Advisor to Silverado Policy Accelerator
• President of Midas Testing Labs
• Active angel investor and hands-on advisor focusing on deep tech, cyber security, and semiconductors. Select recent investments include:

Interests:

• Security of LLMs, and the use of LLMs for Security
• Formal methods, and fault tolerant distributed systems
• Public policy on cyber, semiconductors, and NatSec
• Continuing independent study. Research and application of Deep Learning and LLMs in deep tech. Specifically interested in applying modern techniques to molecular biology. As a bonus, I’m very interested in collaborating on molecular dynamics projects!
• Robotic control systems, simulations (sim2real), and multimodal embodied agents (saycan)
• Multilegged robot and BLDC motor design
• 3D FEM/CFD/FEA/Field solvers. I’m especially curious about the intersection with deep-learning learned-optimizers

Past:

• Staff Security Engineer at Robinhood
• Member of Technical Staff at Meraki
• Developed broadband point to point wideband OFDM modem
• Developed multiple real-time UAV control platforms
• Developed first open source SuperSpeed Software Defined Radio
• Open sourced an IEEE 802.11 compatible VHDL modem – bladeRF-wiphy

Contact:

Please reach out if you would like to discuss any of these topics. Happy to collaborate with researchers, and use my expertise and network to help founders at all stages! Email / Twitter / Google Meet