Suppose someone makes a lengthy computation, and gives you the result. How can you be sure that result is correct without doing the computation again yourself ? The problem raised here is "verifiable computation". If one allows a very short certificates to be produced along with the result of the computation, then it could be doable to check the result much more quickly than redoing the computation.

A natural blockchain application is on chain verification of off chain computation : a ressource constrained smart contract could verify at low cost the correctness of a heavy, off-chain computation, if given the result and a certificate. One approach to this topic originates back to the famous PCP theorem, a decades old very theoretical result in computer science which is now becoming popular and practical in blockchains under the name of "STARKS".

PhD student
Sarah Bordage, advised by Daniel Augot, focuses on new constructions for verifiable computation, using coding theory.


We combine game theory, mechanism design and econometrics to study the state of Decentralized Finance (DEFI) and forecast its evolution.


We build quantitative models that characterize the optimal design and equilibrium pricing of tokens.