-
Computer Science and Engineering student at IST (Instituto Superior Técnico), in Lisbon, Portugal.
-
Passion for SysAdmin and DevOps. Currently running a Proxmox homelab running about 15 containerized applications fully deployed using IaC.
-
Making an MAL CLI using Go to manage my Anime (repo public soon)
-
Making an unified car specs scraper + API using Python (FastAPI + BeautifulSoup + Scrapy)
-
Playing piano for the last 8 years, with a passion for music in general.
Running NixOS: config
Programming Languages
Databases
DevOps & Infrastructure
My MsC's Thesis is DRIFT (Decentralized Reputation Infrastructure for Trust), a novel token-based reputation system designed for decentralized cloud computing platforms governed by Decentralized Autonomous Organizations (DAOs), where its dual-purpose token merges governance power with economic rewards. The system integrates attestation protocols and a stake-weighted reputation model that dynamically reflects verified contributions of computing resources.
Current investigation is focused on how system such as Truebit, Golem, iExec and Akash verify work. Seeing as current decentralized computing networks use TEE (Trusted Execution Environments) as attestation protocols and as a source of truth, DRIFT aims to enter a "Post-TEE" market where, after discoveries such as TEE.fail and CacheWarp, TEEs cannot be considered invulnerable. Thus, DRIFT will use Optimistic Verifications using a Game Theory-focused approach, attempting to have nodes pass a "Bait Attack" where computation results are known a priori and if the node fails the attack, its reputation will be slashed.
To mitigate Sybil attacks, DRIFT will use Hardware-based Identity through PUF (Physical Unclonable Functions). Exit scams will be mitigated via a "Reliability is Blind" approach, where both parties in an exchange will be punished if the task result is disputed and cannot be cryptographically resolved.
The code and algorithm will be open-sourced and it will have a PoC in my GitHub when the dissertation gets released.