7734
Fran Algaba
@fran #7734
co-founder gizatech.xyz 🚀 / prev. Head of ML Eng. @adidas / ZKML / tech+optimism.
109 Follower 92 Following
The future of Web3 is agentic.
Presenting Giza Agents: A modular framework for integrating trust-minimized ML into on-chain strategy and action.
https://www.gizatech.xyz/collection/introducing-giza-agents-framework
Presenting Giza Agents: A modular framework for integrating trust-minimized ML into on-chain strategy and action.
https://www.gizatech.xyz/collection/introducing-giza-agents-framework
Introducing Token Volatility Forecasting tutorial! 📈
Using the complete Giza stack, we predict the volatility of various tokens, compare these predictions against a benchmark, and execute the AI action.
https://t.co/KuLdf2QAmF
Using the complete Giza stack, we predict the volatility of various tokens, compare these predictions against a benchmark, and execute the AI action.
https://t.co/KuLdf2QAmF
Aave Liquidation Prediction Tutorial is live!
Learn how to use Giza Datasets and the Actions SDK to develop a Verifiable ML Model for forecasting the liquidation of debt positions in
Aave V2 and V3 protocols.
https://t.co/LS5gE743XF
Learn how to use Giza Datasets and the Actions SDK to develop a Verifiable ML Model for forecasting the liquidation of debt positions in
Aave V2 and V3 protocols.
https://t.co/LS5gE743XF
Predicting the utilization rate of a given market on Compound Finance V2 🔍
• Data collection, preprocessing
• Building a neural network in PyTorch
• Transpiling, deploying, and running a verifiable inference with Giza CLI
https://actions.gizatech.xyz/tutorials/compound-v2-utilization-rate-prediction
• Data collection, preprocessing
• Building a neural network in PyTorch
• Transpiling, deploying, and running a verifiable inference with Giza CLI
https://actions.gizatech.xyz/tutorials/compound-v2-utilization-rate-prediction
Introducing Token Trend Forecasting 📈
In this first AI Actions tutorial, you will build a neural network for predicting token trends.
We will guide you through a comprehensive workflow using the Giza stack, from data preparation to model training and deployment.
https://t.co/558fCufkZA
In this first AI Actions tutorial, you will build a neural network for predicting token trends.
We will guide you through a comprehensive workflow using the Giza stack, from data preparation to model training and deployment.
https://t.co/558fCufkZA
Relentless shipping 🫡
Calling all ML developers, data scientists, and ZK builders!
Join Gizathon and Build ML Models for Web3, with Giza’s tools for ZKML.
📅February 19th - 26th
🏆$10,000 prize pool
📍Online
Register here: https://764trmbrrb1.typeform.com/to/ApNFyqgm
https://twitter.com/gizatechxyz/status/1755986306625089550
Join Gizathon and Build ML Models for Web3, with Giza’s tools for ZKML.
📅February 19th - 26th
🏆$10,000 prize pool
📍Online
Register here: https://764trmbrrb1.typeform.com/to/ApNFyqgm
https://twitter.com/gizatechxyz/status/1755986306625089550
The Actions SDK is the result of months working closely with web3 protocols, understanding their needs and what's necessary to bring production ready ML to web3 using ZKML.
Our mission is reducing the entry barriers of building ML in web3 so AI talent can easily build in the ecosystem ✨
https://shorturl.at/kHYZ3
Our mission is reducing the entry barriers of building ML in web3 so AI talent can easily build in the ecosystem ✨
https://shorturl.at/kHYZ3
Adoption is about building systems not just the technology
If we want ML to thrive in web3, we need the right tools and products for developers to build ML solutions that create value in protocols.
Giza Datasets makes easy to build ML models for web3 use cases using curated and labeled data.
https://www.gizatech.xyz/collection/introducing-giza-datasets
Giza Datasets makes easy to build ML models for web3 use cases using curated and labeled data.
https://www.gizatech.xyz/collection/introducing-giza-datasets
Building ML for web3 is challenging but there is a great opportunity to bring smart contracts to a new frontier using ZKML.
Very happy to introduce AI Actions as our approach to scale and nurture this synergistic combination with the best developer experience for it.
https://www.gizatech.xyz/collection/ai-actions
Very happy to introduce AI Actions as our approach to scale and nurture this synergistic combination with the best developer experience for it.
https://www.gizatech.xyz/collection/ai-actions
Distributed inference != Decentralized inference network
One is great for scaling production ML systems, the other adds an unnecessary overhead with potential monitoring, performance and latency issues in order to get “decentralization”.
One is great for scaling production ML systems, the other adds an unnecessary overhead with potential monitoring, performance and latency issues in order to get “decentralization”.
2024 is the year ZKML goes to production at scale with deployments in days not months and huge leaps in developer experience⚡️
What's different about ZKML projects? How do you limit risks and build a good solution? New blog post out!
👉 The ZKML workflow
https://www.franalgaba.me/the-zkml-workflow/
👉 The ZKML workflow
https://www.franalgaba.me/the-zkml-workflow/
ML requires fast adaptation for data shifts and model decays. Decentralized networks for inference add an unnecessary overhead in complexity for data observability, model monitoring and ML lifecycles. Critical components for reliable production ML systems.
The struggles that ZKML have for adoption are very similar to the ones conventional ML had almost a decade ago. Let's fix that 🤝
Building ZKML solutions should have the same developer experience and workflows as conventional Machine Learning.
Starting today, I'm rolling out a series of articles on ZKML. I'll be exploring this topic with content that haven't been discussed much. First one is simple:
👉Introduction to ZKML for blockchains
https://www.franalgaba.me/introduction-to-zero-knowledge-machine-learning-for-blockchains/
👉Introduction to ZKML for blockchains
https://www.franalgaba.me/introduction-to-zero-knowledge-machine-learning-for-blockchains/
People really underestimate the complexities of machine learning solutions compared to general software. In the intersection of AI x Crypto is even more critical due to SC interaction.
Decentralized compute or ZKML prover services are not enough for adoption neither production.
Decentralized compute or ZKML prover services are not enough for adoption neither production.
Happy to share I’ll be at the StarkWare Sessions in Tel Aviv!! 🚀
I’ll be presenting gizatech.xyz latest updates in ZKML at the Startup day!
Anyone here going? If so, happy to chat there!! 🙌
I’ll be presenting gizatech.xyz latest updates in ZKML at the Startup day!
Anyone here going? If so, happy to chat there!! 🙌
Really enjoying the vibe here! Content quality in Farcaster is 👌.