Inspiration
I was born in Myanmar and have lived in the United States since I was 17. My parents remained in Myanmar. In 2025, the junta assassinated my mother there. Her death was the direct reason I created Win's VPN.
For people in Myanmar, a VPN is not only a convenience. Authorities censor information, throttle and block connections, monitor networks, and restrict VPN use. A tool that is easy to fingerprint, responds predictably to an active probe, or leaks traffic when a connection changes can create real danger for the person using it.
Win's VPN Inc. is the 501(c)(3) nonprofit I created to work on that problem. The goal is direct: give people a private connection that keeps working on hostile networks without collecting their browsing activity. The same system can help users elsewhere who face surveillance, unsafe Wi-Fi, malicious interception, scams, malware, or targeted network disruption.
What it does
Win's VPN is a no-logs, cross-platform VPN powered by Winn Protocol. It combines censorship resistance, scam and malware blocking, connection failover, and platform-specific leak controls in one client.
Winn Protocol is a custom transport designed for networks where ordinary VPN traffic may be detected or blocked. Its standalone datapath uses X25519 for the authenticated handshake and mixes ML-KEM-768 into the session for post-quantum confidentiality. Gateway identity is distributed through ML-DSA-65-signed descriptors. Unauthenticated probe traffic receives no protocol reply. A fail-closed tunnel gate prevents packet carriage while the session is not ready, and make-before-break routing allows the client to move between available paths without intentionally dropping the existing connection first.
The current pre-release client surface covers major x86 and arm64 desktop and mobile platforms, browser extensions, Apple Watch, Wear OS, Android TV, and Apple TV. Platform protections are implemented separately because a phone, desktop, browser, watch, and television do not expose the same networking controls.
How we built it
The user-facing application is built with Flutter and Dart. The protocol core, session engine, gateway, and hub components are primarily written in Go, with native platform integration where the operating system requires it.
The system includes:
- A standalone encrypted datapath with X25519, ML-KEM-768, ChaCha20-Poly1305, SHA-256, and HKDF-SHA-256.
- ML-DSA-65-signed gateway descriptors and optional dual-attestation support where enabled.
- Silent rejection of unauthenticated probes.
- Bounded replay protection, rekeying, path-MTU handling, and traffic-length shaping.
- Make-before-break routing, gateway failover, and deterministic fallback.
- Fail-closed tunnel state plus platform-specific routing and leak protections.
- Scam and malware blocking without building a browsing-history business model.
- Clients and integrations spanning desktop, mobile, browser, watch, and television form factors.
At a recorded pre-release baseline, 1,503 Dart tests and 6,398 Go tests passed. The software also carried real traffic between physical machines in controlled testing. An internal adversarial review found release-blocking issues, and confirmed findings were remediated with regression coverage. Winn Protocol remains a pre-release production candidate and has not yet completed the independent security audit we are seeking.
Challenges
The hardest part was not simply encrypting packets. It was making failure behavior safe and predictable across very different operating systems.
Hostile networks can block UDP, throttle endpoints, replay packets, classify stable message sizes, or actively probe a server to see how it responds. Mobile devices change between Wi-Fi and cellular networks. Desktop operating systems expose different routing, firewall, and privilege models. Apple platforms impose constraints that differ from Linux or Windows. Browser extensions, watches, and televisions add more seams where a shared assumption can fail.
We also had to separate honest security properties from marketing language. Winn Protocol reduces several fingerprinting and probing surfaces, but no VPN should claim to be invisible or immune to determined traffic analysis. Passing a large test suite is useful evidence, but it is not a substitute for independent review or real deployment measurement.
What we learned
We learned that fail-closed behavior must be designed as a state machine, not added as a final checkbox. Every transition matters: startup, reconnect, gateway change, network change, crash recovery, and shutdown.
We learned that censorship resistance is a full-system problem. Cryptography, packet behavior, gateway configuration, deployment reproducibility, endpoint discovery, logging, and client routing all affect whether a censor can identify or disrupt a connection.
We also learned to treat adversarial testing as part of development. Reviews that found uncomfortable problems were valuable because they produced concrete fixes and regression tests before a broader public release.
Social impact
Win's VPN was designed first for people in Myanmar who may face consequences simply for trying to reach blocked information or communicate privately. The product is also intended for journalists, human rights defenders, civil society groups, activists, and ordinary users facing similar conditions elsewhere.
The nonprofit structure matters. The purpose is not to monetize surveillance or sell browsing data. The planned Shipaton model is a low-cost supporter subscription for users in markets where they can pay, helping fund free or reduced-cost access for high-risk users and continued security work. RevenueCat integration is not complete yet and will be implemented and tested before the final Shipaton entry. We do not plan to place ads inside a privacy-focused VPN.
What's next
Before final Shipaton submission, we will integrate RevenueCat, verify a real purchase and entitlement flow, publish the first eligible store version, provide judges with a free trial or promo access, record a public demonstration, and complete the required store and media assets.
We also plan to publish Winn Protocol under AGPLv3 before an external audit begins, provide a live deployment for bounded black-box testing, remediate independent findings, and publish the resulting technical documentation after responsible disclosure.
Log in or sign up for Devpost to join the conversation.