Dylan Paiton

Crypto data engineering is hard, especially with so many tools and platforms. I've spent years working across the data stack writing pipelines in python, typescript, SQL, and solidity. For my 2025 annual data guide, I'll distill my learnings and approaches to this problem.

Hyperdrive now displays the Variable APY and capital multiplier directly on the Pools page. This means if you open a short with 1 ETH, you get 67.2 ETH worth of exposure to Lido’s staking yield for 6 months. @hyperdrive_box @HyperVueFDN @delv_tech

The fixed-income primitive for DeFi Hyperdrive is a decentralized protocol that enables a truly organic, fixed-income-powered market where participants can engage in opportunities such as: • Purchasing fixed-rate positions by longing. • Opening short positions for leveraged variable-rate exposure. • Providing liquidity to earn from three different revenue streams.

Data sovereignty is only possible on Ethereum. It goes beyond ideology. "Running Big Query on your laptop" with OSS tools is faster cheaper and let's you do more with less. @notnotstorm's talks are an inspiration to watch. link below

Many people arguing about blobs, but so far no one has simulated how they respond to demand... until now. TL;DR: Blobs are insanely bullish for ETH long term

@fubuloubu @mihai673 100% agree -- pypechain is a way to automagically interface with the core functions of the contracts in a way that provides for a much improved developer experience over web3py. It doesn't do any of this for you, that all goes on top.

@mihai673 ... Even a tenth of the battle. To call it right, you need to do path indexing to find the most liquid paths to submit your trades through. Then you need to do simulation logic on top of that to make sure that the parameters which it shows is correct and okay.

Been poking around some of the "AI frameworks" for crypto past few days, have noticed most are fairly limited to picking from one of a few options in their SDK, and a few only deploy on one chain Silverback has no limits. Listen to any event and sign any txn with AI on any chain

@fubuloubu @mihai673 @lundquistsheng I could see a caching system working well; that's a good idea. We'd love to chat more about how you could use pypechain. We have a list of features we're jamming on presently & if we find that there's mutual interest in some of them we can adjust priority and exchange resources.

@fubuloubu @mihai673 @lundquistsheng You have to compile (transpile?) the abi into python code. We do this and auto-upload it to pypi on build for Hyperdrive (pypi.org/project/hyperd…), which includes all of its dependency smart contract abis and (currently) most of the integrations.

According to Franklin Templeton, clearing tokenized assets is over 30,000x cheaper. Yes, you read that right. On stage at the RWA Summit, Head of Digital Assets Roger Bayston compared the total cost of clearing 50,000 transactions: Legacy system: ~$50,000 Blockchain: $1.52

The Python ETHcosystem (heh) continues to grow, and @notnotstorm has been a big part of it! Looking forward to the talk.

new example: auto-redeem withdrawal shares

Fixed borrow is a super interesting financial tool that will be useful for many types of people. Stable borrow rates make it easy to maximize your return by leveraging yield-generating assets.

Compete for prizes & get hands-on support for building python tools to automate Hyperdrive trading, build data science pipelines, and more at @ETHGlobal hackathon! Details on what we're hacking on are here: youtu.be/_GS1x2OSbwE

“Ok, fine, but why wouldn’t I use an if/else statement instead?” Write up all of these examples with if/else and you’ll see. It can be cumbersome, it does not trivially support the Result and Option syntax, it’s boomertech. Get with it.

Or you can flip the sign of an integer:

You can use them for mapping numerical values from one range to another. We use it here when mapping a signed integer to an unsigned one:

Match statements are great for tests. I often nest them to catch errors & panics.

Ok so this one is a classic, albeit controversial on stack overflow. I actually hated it for a while, but I like it now. It's more "code looking" but also more complete for many conditions w/o thinking of the "else" error condition.

Running an iterative loop that should end early on success? match!

I feel compelled to shout out match statements in Rust. They’re so great. Especially for error handling, but also for so much more. The Rust docs have a ton of great examples, but I wanted to highlight ones we actually use in our codebase.

Need to handle optional inputs? match!