Doron Zarchy

Computer scientist and Applied Cryptographer at Fhenix, working on FHE, threshold cryptography, zero-knowledge systems, and verifiable democratic technology. Direct democracy enthusiast.

PhD, Hebrew University of Jerusalem

Research Associate at APSIA

Research

Aggregator-Based Voting using Proof of Partition

IACR ePrint 2026/545 · ACM ASIACCS 2026

A voting protocol built around proof-of-partition ideas, aimed at making aggregation verifiable while preserving the privacy and structure needed by practical election systems.

High-Precision Exact FHE Made Simple, General, and Fast

IACR ePrint 2025/2321 · accepted to CRYPTO 2026

Work on exact high-precision fully homomorphic encryption, focusing on practical techniques that make encrypted computation simpler to implement and faster to run.

High-Throughput Universally Composable Threshold FHE Decryption

IACR ePrint 2025/1781 · ACM CCS 2025

Threshold decryption for FHE systems with a universal composability treatment and an implementation-oriented emphasis on throughput.

Projects

Subsenate

GitHub →

A zkSNARK-based e-voting system that randomly selects a community of judges from a pool of eligible voters — cryptographic sortition. Using zero-knowledge proofs, the selection is verifiably random and private: nobody can predict or manipulate who gets chosen, yet anyone can verify the result is fair.

"Politicians and diapers must be changed often, and for the same reason." Subsenate takes the turnover problem seriously: sortition — random selection of citizen decision-makers — powered Athenian democracy, and ZK proofs can make it verifiable today.

Coercion-Resistant Elections

A full Rust implementation of coercion-resistant e-voting based on the JCJ protocol and Civitas. Implements the complete cryptographic core: DVRP credential issuance, threshold ElGamal key generation, PET-based duplicate and invalid-credential elimination, a two-pass RPC MixNet with public verifiability, and distributed decryption — all wired into a runnable election pipeline with a web UI.

The key insight of JCJ: a coercer can force you to vote under threat, but cannot prevent you from casting a silent override later. Your real vote always wins.

The Social Evidence

A browser extension for collaborative fact-checking. Select any text on a news page, attach a comment linking it to contradicting evidence, and every visitor with the extension sees the same annotations — turning passive reading into a shared, persistent record of public inconsistency.

Built around the idea that accountability needs infrastructure: not just the claim that a politician contradicted themselves, but a linkable annotation anyone can see on the original source.

Aggios EPA Benchmark

GitHub →

Benchmark code by Marius Lombard Platet for evaluating Aggios on EPA-style workloads, built to make the system's performance claims reproducible rather than anecdotal. The repository collects the scripts, inputs, and experiment scaffolding needed to compare configurations and understand where the implementation spends its time.

This kind of benchmark is the bridge between a cryptographic protocol and an implementable system: it turns asymptotic or architectural claims into measurements that other people can inspect, rerun, and challenge.

On Democracy

To me, democracy is more than a form of government. It is a way of giving people real influence over the decisions that affect their lives. It is also delicate: easy to undermine, and much harder to design so that it is fair, resilient, and trustworthy.

This challenge has motivated much of my work in cryptography. I am interested in how cryptographic protocols can strengthen democratic institutions by reducing the need to trust the people who operate them, while preserving privacy, fairness, and public verifiability.

I have explored this idea from several directions. Aggios is a protocol I co-designed to make large-scale direct democracy more practical. It is built around a new zero-knowledge proof primitive called a proof of partition, which allows citizens to participate in frequent decisions while preserving privacy and public verifiability. Subsenate is a sortition system based on zero-knowledge proofs that randomly selects citizen judges from a pool of eligible participants — an ancient idea implemented with modern cryptography. I also built CRelections, a complete implementation of a coercion-resistant electronic voting system, demonstrating how advanced voting protocols can be turned into practical software.

The question that motivates me is simple: how can we build democratic processes that people can trust, even when they do not trust the authorities running them?

Contact me

Feel free to reach out about research, collaborations, or just to say hello.

doronz@gmail.com