ElectionChamp

E-voting platforms are typically hardened against outside attackers and surprisingly open on the inside. An administrator who can watch results arrive in real time can leak momentum, apply pressure, or simply undermine a vote's credibility.

Years in the online-election space convinced me the industry had stalled on exactly this gap, so I built ElectionChamp to close it: voter privacy enforced by how the system works, not promised by policy.

I was the sole engineer and designer for the whole ecosystem: the voting logic, the encryption, the admin dashboards, and the public marketing site. AES-256 encryption, full audit trails, and a voter experience simple enough that running an election does not need a manual.

Every voter receives a unique sixteen-character high-entropy key instead of a reusable login. The moment a key is used to cast a ballot it is invalidated, so it can never be used twice. Ballots record IP and timestamp for post-election forensics without ever linking a ballot to a voter's identity.

The system deliberately hides the voting status of the most recent voters and randomizes the display order, so an administrator can never deduce who cast the latest ballot.

Every administrative action is itself logged, and unusually frequent result-viewing is flagged in the audit trail. Watching the watchers is a feature, not an afterthought.

Ballot setup is the biggest administrative bottleneck, so I automated it with a schema-enforced pipeline. An organizer describes the election in plain language, a board of directors vote for example, and the AI returns a validated JSON structure that renders directly as a working, logically sound ballot. Structured output, not open-ended chat, is what makes generation dependable in a compliance-shaped product.