The problem Memeswap solves

Memeswap provides audited token contracts for meme deployers to ensure streamlined meme launches with transient rental liquidity, generating yields on top of meme traffic with its built-in ETH vault and gamification aspect. The platform allows users to launch and trade memes while also offering an opportunity to generate safe and substantial yields for stakers through the ETH vault.

Memeswap is not just another token launcher with a bonding curve. Instead, it focuses on creating a capital-efficient environment leveraging meme traffic. While emphasizing efficiency and yield generation, it also offers social features similar to profiles and chatboxes on the platform.

Challenges I ran into

1. Serverless billing:

While building the frontend application, we prioritized simplicity and ease of hand-off so we decided to use Vercel’s managed databases(postgre & redis) for user and token data and chat functionalities. During internal tests with few things going, the bill reached $5. We realized this is not so scalable after all and decided to ditch Vercel completely. Our solution was; a “baby swarm” on a dedicated VPS fronted with Cloudflare. This way we had limitless horizontal scaling via Docker swarm, security and caching via Cloudflare and flexibility to add services on-the-go. We also build a webhook to support continuous deployment via github actions. No need for Vercel now :)

2. IPFS pre-determined CID

In token launches, we write IPFS CID of token image uploaded into the token contract itself. Problem arises from pre-determining the CID of token image since we need to make sure a token is deployed before we pin the image and keep it pinned(we pay for this). Initially, we assumed SHA256 digest of the image produces the CID. It worked for some images and didn’t for some images. So we went on to investigate and found out it was the chunk size that’s creating the problem. IPFS chucked files into 262144 bytes before recursively hashing the tree so our solution stopped working at larger files. Solution was simple after this point. We implemented the official node library and used their customizable CID generator to use IPFS’s chosen size. Now we have future-proof CID’s in token contracts.