Skip to content
View past5's full-sized avatar
♾️
creating and building
♾️
creating and building

Highlights

  • Pro

Block or report past5

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
past5/README.md

Sev Geraskin

Co-Founder & VP Engineering, PolarGrid · President, Economy of Wisdom Foundation · Vancouver, BC


Real-time AI inference at the edge.
Distributed systems that scale.
Frameworks that ask whether technology actually serves people.

What I'm Building

PolarGrid — North America's first real-time AI inference compute network, purpose-built for voice agents and agentic AI workloads. Sub-30ms latency. Edge-first architecture. Bare metal Kubernetes, Triton Inference Server, and a full voice pipeline (Whisper → LLM → Kokoro) running on Blackwell GPUs at colo in Vancouver and Toronto.

Economy of Wisdom Foundation — A research and community organization developing new frameworks for post-productivity economies. The core question: Does a system increase or decrease fulfillment for the people it touches?

Past 5 Solutions — Founder. Data storytelling and technology strategy consulting.


Work That Mattered

  • Scaled distributed data services 100× at Mastercard while cutting cloud costs 75% — architected pipelines that unlocked multimillion-dollar opportunities
  • Built PolarGrid's full infrastructure stack from scratch: two-tier Kubernetes, GPU passthrough, MetalLB, bare metal Supermicro
  • Mentored four AI startups through Scale AI's Zen Launchpad Cohort 2
  • Published Transmutarianism (DOI: 10.5281/zenodo.18809258) — a moral accounting framework for measuring deprivation and fulfillment flows through agent networks: M = τF + A
  • Developed PITC (Physical-Information Theory of Consciousness), distinguishing crystallized intelligence from fluid consciousness in AI-integrated substrates

Technical Depth

Infrastructure     Kubernetes (MicroK8s, Kata containers) · MetalLB · bare metal GPU clusters
AI/ML             Triton Inference Server · Whisper · Qwen · LLaMA · Nemotron · Kokoro TTS · NVIDIA Blackwell
Cloud             AWS (Solutions Architect, Security certified) · multi-region distributed systems
Languages         Python · Go · TypeScript · Bash
Domains           Voice AI · agent orchestration · AI alignment · real-time inference · LLM serving

Thinking

I've spent years asking whether the systems we build actually serve people: as a measurable, falsifiable question. Transmutarianism is my attempt to formalize that: a framework for tracking whether agents (human or artificial) are net producers of fulfillment or deprivation across the networks they touch.

I believe compute infrastructure is a question of cognitive sovereignty. Who controls the inference layer shapes who benefits from AI.


Credentials

BCIT · AI Alignment · AWS Solutions Architect · AWS Security · Generative AI · PMP


Connect

LinkedIn · PolarGrid · Economy of Wisdom


The real measure of any system is whether it transforms deprivation into fulfillment for the people and communities it serves.

Popular repositories Loading

  1. chat-gpt-prompt chat-gpt-prompt Public

    This repository contains Python scripts of prompt examples taken from Isa Fulford and Andrew Ng fantastic ChatGPT Prompt Engineering Course

    Python 2

  2. WashBench WashBench Public

    Summarization Relevance Benchmark for Large Language Models

    Jupyter Notebook 2

  3. cron-email cron-email Public

    a shell script that schedules cron jobs to send emails with attachment

    Shell 1 1

  4. ai-marketing-slogans ai-marketing-slogans Public

    Marketing Slogans Created by using Adversarial Approach with GPT and Claude Agents

    1

  5. we-the-ai-assistants we-the-ai-assistants Public

    Studying Emergent In-Group Behavior in Large Language Models via Adversarial Agent Instructions

    Jupyter Notebook 1

  6. universaltheoryofconsciousness.github.io universaltheoryofconsciousness.github.io Public

    The Universal Theory of Consciousness

    1