STARKs are a promising new direction for zero-knowledge proofs. It is currently being commercialized for an application in cryptocurrencies, yet the approach is very attractive in general to solve a number of generic privacy-related challenges, albeit it comes at significant computational costs.
What is needed at its core is very often the proof of knowledge of a preimage of an output of a hash function without revealing said preimage. Currently used hash functions are not at all designed for this purpose, and we expect that with custom designs overall protocol performance can be spend up by more than an order of magnitude.
The goal of this project is to design and analyze such a hash function, and together with this develop a suitable use-case.
Taiga RIG Inventory project here