Every year in America, there are approximately 463,634 victims of sexual assault. However, this number is projected to be much greater, as 75% of sexual assault cases go unreported. The reason why sexual assault often goes unreported lies in the fear and risks of coming forward. Victims often fear that no one will believe them, that it is their fault, that the issue is “not serious enough”, that their reputation will be ruined or they might be subject to defamation lawsuits, or just want to forget it ever happened because filing a formal report is often retraumatizing. As such, their options for coming forward safely remain limited, which perpetuates the issue as perpetrators are frequently serial offenders and rarely face justice. Though staying silent is the “easier” option, it creates a false sense of security for the victim, as it only empowers the perpetrator and increases the number of serial offenders.

Solution

Zuvivor is a platform that leverages programmable cryptography & knowledge graphs to empower sexual assault survivors to anonymously report their abusers in a safe and secure environment that eliminates first mover disadvantage and related fears that come with speaking out. It allows them to create a timestamped record of their assault with secured claim filing/evidence collection and be matched with victims of the same perpetrator so they know that they are not alone on this journey and can help one another find a path forward together.

Zuvivor (1).png

How it Works

Report - Victims submit reports by completing a form where they fill out identifiers about their perpetrator (i.e. name, socials, physical descriptors, image), incident info (i.e. what happened, can include screenshots or receipts), and verify their identity via KYC, ZKEmail, or WorldID. This ensures that the report was submitted by a real person to maintain the integrity of the platform being a safe space for victims. These reports are also timestamped and stored as transactions on-chain with Arbitrum. I also previously successfully implemented GPT-4o to be able to describe an uploaded image with text to simply machine readability.
Cryptography - Building on Cursive's technology, the system uses MPC (Multi-Party Computation) and PSI (Private Set Intersection) to identify overlapping information in the reports submitted from different victims without revealing anything else about their identities to each other.
Knowledge Graphs - Networks of perpetrator/victim nodes develop from overlapping data in the edges which hold information about the case and signify a perpetrator-victim relationship. Clusters will eventually start to form around specific perpetrator nodes as more information comes to light from different victims coming in with similarities but also different pieces of the puzzle, and the victim nodes may gain access to the network of other victims in the cluster. This structure ensures against definite matching simply based on name (multiple people with the same name) or requiring that the victim knows a unique identifier (e.g. email).
Match - Once victims become part of a cluster, they can check their perpetrators to view number of reports filed against them and match with other victims of the same offender. The platform notifies victims when these matches occur, offering them the option to connect in an anonymous, private chat. Within this secure environment, survivors can decide if and when to share personal information, empowering them to connect and find support while preserving privacy and control.

New at ZuThailand

A Graph Neural Network (GNN) architecture using PyTorch for the matching algorithm.
Victims can view the number of commonalities they have with one of their matches by generating ZKPs (using EZKL's API) on the different fields, but maintaining privacy of which fields these are.
Because the matching algorithm is a public model trained on sensitive, private data from victims, EZKL can help to prove that the benchmarks are true for the committed model and private data, without a reviewer needing to access that data. This way, we can prove that the model is acting in good faith (i.e. matching victims of the same perpetrator together to the best of its ability).

Tech Stack

I built this using Next.js for the frontend, Flask for the backend, and PostgreSQL for the database. ZKEmail and WorldID were used in the verification step of the form. Arbitrum Stylus was used to write the smart contract in Rust. EZKL was used to easily generate ZKPs.

Traction

Zuvivor has garnered widespread support from various communities committed to creating safer spaces, including people in fields like modeling, martial arts, and more who are eager to share the project with their members.
Additionally, Zuvivor has gained significant traction on social media, particularly on crypto twitter in response to the news regarding DWF and other related discussions dedicated to amplifying awareness and the mission. This strong, cross-community interest underscores Zuvivor’s potential impact in fostering connection and healing for survivors across diverse spaces.
Zuvivor ended up winning 4 prizes at the Edge City Hackathon, including 1st place Arbitrum Innovation Track (for use of Stylus), community choice (chosen by whichever finalist project received the loudest cheer from the crowd), World ID pool prize, and Deshittification Residency recognition.
Right after the hackathon awards ceremony, I was interviewed by Arbitrum on winning 1st place, and the video was widely reposted on Twitter. People even began sending funds to support the project!
The following week, I pitched Zuvivor during DevCon to various individuals and was surprised to hear that they had already heard about the project, but didn't know I was the one behind it. I also got introduced to different C-suite executives of well-known companies who began talking about the project in their internal channels and fast-tracking me through their grant process.

Future Steps

I'm currently in conversation with different ecosystems who are seeking to provide mentorship and guidance, and actively applying to different grant opportunities. I'm also looking for a technical cofounder to join me in making this a reality by ETHDenver 2025.