Sawtak (صوتك) – Decentralized Public Complaints Platform
Tamper-proof public issue reporting platform with cryptographic anonymity and state machine consensus.
What was broken before this system existed?
Traditional public complaint systems suffer from centralized censorship, lack of verifiable audit trails, and citizen fear of retaliation. Citizens needed a cryptographic guarantee that their reports could not be quietly deleted or modified in a database, while authorities required verifiable audit trails.
What did I actually engineer and own?
As the Blockchain & Systems Engineer, I owned the system end-to-end: from initial requirements definition and architectural design to database schema modeling, backend pipeline development, and final production deployment.
Architected a dual-layer system using Cosmos SDK and CometBFT (Tendermint Core) consensus. State transitions (complaint submission, authority status updates, resolution proofs) are permanently recorded on an application-specific blockchain. Coupled with an IPFS decentralized storage pipeline and a high-speed Bun/PostgreSQL CQRS read cache for public exploration.
How the system is structured
Two-Layer State Management: Architected a clear separation of concerns where CometBFT handles immutable block storage while an embedded LevelDB manages mutable application state for rapid access.
Decentralized P2P Network: Deployed a 3-node validator network communicating via an authenticated Gossip Protocol over port 26656.
Containerized Infrastructure: Containerized the entire blockchain network using Docker and deployed it on a Hostinger VPS to simulate a realistic production environment.
High-Availability Indexing: Built a custom Cosmos Indexer service using Bun/Elysia that polls the CometBFT RPC endpoint, parsing block events and synchronizing them into a relational PostgreSQL database to facilitate high-speed, complex frontend querying.
Why this architecture? Tradeoffs & judgment
1. Why Cosmos SDK / Sovereign App-Chain over Smart Contracts (Ethereum/Solidity): App-chain provides custom state transition rules, zero gas fees for citizens via fee-grant modules, deterministic block finality (CometBFT), and full sovereignty over governance.
2. Why Two-Layer State Architecture (CometBFT + PostgreSQL CQRS): Blockchain is the single source of truth for write transactions; PostgreSQL acts as a read-model cache populated by real-time blockchain indexers to enable millisecond search queries without hammering validator nodes.
3. Why Ed25519 Ephemeral Keypairs: Allowed zero-knowledge style unlinkable complaints while providing citizens with a cryptographic receipt to check status without a persistent login identity.
4. What happens at 10x traffic: CometBFT transactions scale horizontally with pipelined mempool gossip; state pruning ensures validator disk usage remains bounded by archiving historical state to decentralized storage.
5. What was deliberately NOT built: Did not implement volatile speculative tokens or DeFi mechanics; focused strictly on cryptographic verification and civil utility.
Hardest engineering hurdles
Balancing citizen anonymity with spam prevention on a fee-less public chain. Solved through cryptographic challenge-response rate limiting and proof-of-work puzzles at the gateway layer before mempool inclusion.
Custom Application-Specific Blockchain: Engineered from scratch using the Cosmos SDK framework and Go, rather than relying on external smart contracts.
Proof of Authority (PoA) Protocol: Developed a custom module to restrict block production and validator access strictly to pre-authorized entities, disabling public inflation and staking.
On-Chain Cryptographic Verifier: Implemented logic within the blockchain keeper to verify Ed25519 signatures, authenticating anonymous complaints mathematically without exposing PII (Personally Identifiable Information).
Security Middleware: Designed a custom AnteHandler decorator to intercept transaction pipelines, effectively mitigating unauthorized state modifications and Authz bypass vulnerabilities.
Decentralized Storage: Integrated IPFS via Pinata to pin evidence files, storing only lightweight cryptographic hashes on-chain to prevent state bloat.
Quantified results & production impact
• Engineered a 4-node local testnet achieving deterministic sub-2s block finality with CometBFT consensus.
• Implemented Ed25519 ephemeral cryptographic receipts, allowing citizens to verify issue progress without maintaining a trackable identity.
• Decoupled blockchain consensus from query serving via CQRS, delivering sub-10ms UI query latency.
What I would change at 10x scale
Separating the write path (consensus state machine) from the read path (CQRS relational projections) is essential for any high-performance distributed application.