Skip to content
View SCJedi's full-sized avatar
0 followers · 3 following

Block or report SCJedi

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SCJedi/README.md

Eric L.

Systems Architect @ Infinite Visions AI Agents — a 3-person AI consultancy.

What I Do

I design AI automations and integrations for businesses, with a simple rule: if the system won't pay for itself, we don't build it. The work spans applied ML research, autonomous agent architectures, and production tooling — whatever the problem actually requires.

Featured Work

Entropy-Adaptive KV Cache Compression

Per-head entropy-based budget allocation for LLM inference. 2.6x BLEU improvement over uniform strategies at 2x compression; extended to 12x lossless compression on a consumer GPU (RTX 3060, 12GB). Python PyTorch HuggingFace Information Theory

Agent Marketplace

Decentralized AI content protocol with Bitcoin-style peer discovery, verification, and reputation scoring. No central directory. Live on a public node. Node.js SQLite Docker P2P

YesBot

Auto-approver for Claude Code's permission prompts during autonomous development sessions. Small tool, big workflow unlock. Python Flask WebSocket

Meal Planner

Weekly meal planning app with OCR recipe import via Tesseract.js and Firebase sync. Built it because I needed it. JavaScript Tesseract.js Firebase

The Philosophy

Every engagement starts with math: cost to build vs. output gained. If 1+1 doesn't equal more than 2, we'll tell you. The best AI consultancy is one that sometimes talks you out of buying AI.

Get in Touch

Business inquiries: Infinite Visions AI Agents LinkedIn: Connect with me (update this link to your actual LinkedIn URL)

Pinned Loading

  1. entropy-adaptive-kv-cache entropy-adaptive-kv-cache Public

    Per-head entropy-based KV cache compression for efficient LLM inference. 2.6x better than uniform strategies at 2x compression.

    Python 3 1

  2. BitNet BitNet Public

    Forked from microsoft/BitNet

    Official inference framework for 1-bit LLMs

    Python 1

  3. yesbot yesbot Public

    Autopilot for Claude Code. Auto-approves, blocks, or escalates tool calls based on your rules. Live dashboard included.

    Python 1

  4. agent-marketplace agent-marketplace Public

    Search engine and marketplace for AI agents — API-first, built for agents to discover agents

    JavaScript

  5. meal-planner meal-planner Public

    Touch-first dark SPA meal planner with AI recipe import, OCR, and Firebase cloud sync

    JavaScript

  6. tonbistudio/turboquant-pytorch tonbistudio/turboquant-pytorch Public

    From-scratch PyTorch implementation of Google's TurboQuant (ICLR 2026) for LLM KV cache compression. 5x compression at 3-bit with 99.5% attention fidelity.

    Python 806 103