SVRGN Weekly Digest #32 💫
AI, axons, explosive birds and chips
🤼 People
Shivoh Chirayil Nandakumar - submitter of PhD thesis and founder at RideScan
Shivoh is one of the rare cases where it discoveries made during PhD-studies directly translate into a potential business advantage and founder opportunity. Without disclosing too much, reliability and safety of autonomous systems is very difficult to make general statements about. Shivoh may have solved this.
💼 Portfolio jobs board
This weeks selection of opportunities from the portfolio:
🚀 Companies
Munin - Portable C-UAS systems
Status: Bootstrapped
Source: Hackathon
Founders: Magnus F. (ex-Stanford, ex-Paratrooper)
Why it’s cool:
New attack methods require new defenses. FPV drones are killing more people than artillery in Ukraine, and no one has a good defensive system. Shotguns and Zeromark are skill-based and hard to scale up. However, shooting tiny guided missiles at them might be a solution.
Tracer Bio - Observability for bioinformatics
Status: Angel-round 3x oversubscribed
Source: CAVI
Founders: Laura Bogaert, Vincent Hus
Why it’s cool:
Sure, bio-data is the largest growing data-category in the world, but that doesn’t mean we have nice ways to handle that data. If you’ve studied biology (or have biology-trained family members who you have to help with computers) you might guess stuff like git, parallel computing and memory management aren’t part of the curriculum. Tracer might be solving this.
PolyN - Neuromorphic Front-End chips for sensor data preprocessing
Status: Unknown
Source: CAVI
Founders: Alex Timofejevs, Boris Maslov
Why it’s cool:
Neuromorphic chips can be a lot more energy efficient than traditional von Neumann-architectures, usually by placing the memory closer to the processing. Moving data about is still >60% of all energy used by a chip.
💡 Ideas & Science
IEEE Spectrum - Negative resistance in Neuromorphic devices
Axons (think your way to them) can carry signals insanely well, even blue-whale ones that are 30m long. That’s because they’re actively amplifying the signal on the way, it’s an active transmission line. With the help of the inventor of the memristor, researchers have now been able to reproduce it in inorganic matter. Needless to say: if you’re building this, reach out!
Nothing like some hard grokking to kickstart your brain. “Don’t model bulldozers with quarks” is the quote from physicist Leo Kadanoff, meaning that there are emergent phenomena from constituent parts which aren’t necessarily possible to explain using the properties of the parts. Or at least it’s not efficient to do so.
Breakthrough Technologies for Mineral Exploration
There are a number of technology leaps that have happened in the last few years that drive both the demand side of minerals (gib rock, rock so good for battery), and supply (intelligently scans the horizon for rocks). Someone should look into what one could do with all this…
It’s rumored to be just thinking a bit harder (or CoT), and it magically works! If you needed a sign that inference will rule the world, this is it. Imagine having one inference run for 7.5 million years to respond to a crucial question!