Semiotic Labs May 2023 Update

Executive Summary

In the past month, we have made significant progress in various areas, including the development of the end-to-end prototype for Verifiable Payments/TAP, exploring options for Verifiable Extraction, and releasing v2 of the Allocation Optimizer. Our team has also grown with the addition of a new full-time backend developer. As part of our outreach efforts, we participated in ETHGlobal Tokyo and demonstrated a full-stack web3 demo created with GPT-4. In the upcoming month, we will focus on completing the TAP end-to-end prototype, further developing Verifiable Extraction, and enhancing the Allocation Optimizer, as well as continuing our engagement in industry events and workshops.

Looking Back (Deliverables)

Verifiable Payments/TAP

End-to-end prototype

  • Polished TAPManager/adapters API
  • Implemented TAPManager
  • Implemented Gateway Aggregator
    (Note: Related code will be made public in May)

Security

  • Reviewed dependencies for vulnerabilities
  • Reviewed attack vectors

Verifiable Extraction

  • Semiotic cryptographers met with Alex/StreamingFast to discuss details of Firehose
  • Hosted Dionysis Zindros for presentation at Semiotic Labs office; Dionysis is a Stanford blockchain researcher with lightweight refereed game-based approaches to EVM light client verification, potentially applicable to our Verifiable Extraction efforts

Indexing

  • Firehose mainnet sync complete!
  • All Firehose components added to our k8s
  • Now using Firehose on our indexer (events only)

Allocation Optimizer

  • Released v2 of the Allocation Optimizer. GitHub changelog is here. Notable changes include:
    • Various UX improvements, including config-driven saving data to CSVs to minimize GraphQL queries, binaries
    • Reduced runtime from 1 s to .001 s
    • Improved documentation

Staffing

Outreach

Looking Ahead (Upcoming Priorities)

Verifiable Payments/TAP

  • Develop collateralization smart contract
  • Release TAP public code
  • Complete end-to-end prototype

Verifiable Extraction

  • Finish orientation with Firehose data extraction and its flat-file output consumption (i.e., downstream querying)
  • Complete collection of Verifiable Extraction requirements
  • Complete draft of Verifiable Extraction requirements and possible solutions

Allocation Optimizer

  • Address open support issues and PRs
  • Publish blog post part 2; see part 1 here
  • Enhance logging

Outreach

  • On May 5, Tomasz Kornuta, our Head of AI, will lead an in-person workshop at the E&N House of Web3 on training neural networks for image recognition and text classification; more details here
  • On May 5, Sam Green, our Head of Research, will present at the virtual Coinbase ML and Blockchain Research Summit, discussing early experiment results using GPT-3.5 and GPT-4 to build GraphQL queries; more details here
2 Likes