$300M lost in scam & stuff, human labor is not scalable. ML automates a hell out of it.
-They gathered 40 cluster data, 96% accuracy.
-96% due to overfitting, still need more data. How do you get enough data?
-Scanning entire chain not scalable for Securitas. They plan on building an data augmentation system. It breaks code into preliminary building part and do communicative testing with them.
-Also go to audit company to get data.
-V1 & V2 just a data phase. V3 they start becoming competitive to auditing firms (way faster and cheaper), at V4 looking at taking over human audit.
-V1 2-4wk
-V2 3-4wks
-V3 Not sure on technical level, data sparse here mainly feeding the ML side
-V4 ?
ML vs Formal Verification
-It seems like they don’t know about formal verification
-Non of them are doing it to the extent of replacing auditing firm
-ML can generalize to better than human level
Edge against competitors
-Passion; Robert has done 4 years of ML at the age of 17; Built a Aurras.
Problems
-Contracts want names on website: Certik, OpenZeplin. How do you gain recognition in it?
-In V2 we let ppl use it and get momentum and words out. Interact with auditing firms directing -> they are serving all those auditing firm.
NOte
-Will issue non-transferable NFT. Prob also doing in ICO/equity
-But will be in 1 months to do so
-Currently only two ppl on the team
-Thinking about a VSCode package that prevent weakness during writing
-haven’t looked into automation competitor; only looked at human labor
Want to print your doc? This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
Deus Ex Securitas