TradeLayer

The problem

Institutions want to hold tokenised stocks, but on a public chain everyone can see what they buy and how much. That is a deal-breaker for anyone managing real money.

What I built

Orders are encrypted in the browser and only opened inside a secure enclave, so neither my backend nor the chain ever sees them in plain text. The enclave checks each order against the institution's rules, places it with a real broker (Alpaca, paper trading), and settles trades in netted batches so individual orders don't show up on-chain. Every token is backed one-to-one by a share at the broker.

On the chain side there are five Solidity contracts on Hedera testnet, covering wallet registry, compliance, a pooled vault, escrow and a private ledger, plus tokenised equities with KYC and freeze controls.

The hard part

My first version let the backend read every order. That defeated the whole point, so halfway through the hackathon I threw it away and rebuilt the core around the enclave.

I also hit a bug in Chainlink's beta: confidential HTTP responses carried per-request headers, so the nodes never agreed on a result. I measured it, worked around it, reported it to Chainlink, and wrote the limitation into the docs.

Result

Built solo in nine days, with 87 Foundry tests (12 of them proving one institution can't see another's data) and a written threat model that says plainly what is trusted and what isn't.