Skip to content
View arterialist's full-sized avatar
🧠
Building Artificial Life
🧠
Building Artificial Life

Block or report arterialist

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
arterialist/README.md

Andrew Gostishchev

Senior Fullstack, Backend & AI Systems Engineer with 9+ years of experience building high-throughput distributed systems, real-time voice AI infrastructure, and production machine learning platforms. Based in Kuala Lumpur, Malaysia, and globally available for remote roles.

Currently CTO at AIMA Labs and mcphero.app, and creator of open-source projects including FlexiBilling, unfolder, magic-prompt, and ardkit. My engineering focus spans low-latency voice platforms, autonomous agent tooling, billing engines, and published research in computational neuroscience.

WebsiteLinkedInEmail


Active work and open source

Voice AI and agent tooling

  • AIMA Labs. Voice agent platform running on WebRTC, Gemini Live, and telephony trunks. Voice turn latency stays under 800ms on production PSTN calls.
  • MCPHero. Compiler that turns API specs, cURL commands, and web interfaces into sandboxed Model Context Protocol (MCP) servers in under 30 seconds.
  • unfolder. Deterministic code synthesis harness combining AST tree schedulers and test verification loops with small LLM layers for semantic fills.
  • magic-prompt. Codebase-aware prompt engineering CLI that extracts AST context and expands rough user prompts into structured specs via fast LLM inference.
  • ardkit. SDK in Python and TypeScript implementing the /.well-known/ai-catalog.json standard so AI agents find and route tools dynamically.

Billing and usage metering

  • FlexiBilling. Provider-agnostic usage metering and balance deduction engine. Handles multi-asset priority waterfalls, idempotent transactions, tiered pricing, and double-entry ledgers with matching math across four language implementations.
    Packages: Python (PyPI) • Go (pkg.go.dev) • Rust (crates.io) • TypeScript (npm)

Computational neuroscience and embodied AI

  • Towards Artificial Life (Papers). Author of published research on brain-inspired neural architectures, including PAULA (local learning without backpropagation) and the ALERM framework for biologically constrained spiking systems.
  • Live C. elegans Simulator. Simulation of the 302-neuron connectome and 13-segment body physics in fluid viscosity, streaming over WebSockets at 60 FPS.
  • Neuron Model. Reference implementation of the PAULA predictive coding and spiking neural network architecture.

Selected past projects

  • Callegram. Event management and booking platform inside Telegram. Won first place in the global Telegram Mini Apps Developer Contest (2023).
    • Built two-way Google Calendar sync and conflict prevention with PostgreSQL isolation levels and Redis locks.
    • Developed full-stack architecture with React, Vite, FastAPI, and async SQLAlchemy handling high-concurrency booking traffic.
    • Implemented responsive interface matching Telegram native styling across iOS, Android, and desktop.
  • TheTonpay. Payment gateway and merchant tooling for the TON blockchain.
    • Authored Store and Invoice FunC smart contracts with low-level Bag-of-Cells (BoC) binary cell serialization for non-custodial checkout.
    • Built merchant client SDKs in TypeScript, Python, and Go with HMAC-SHA256 webhook validation and QR payment generation.
    • Created reference Telegram Web App commerce checkout with instant wallet settlement via TON Connect 2.0.
  • Pokemod. Android automation platform and device management infrastructure.
    • Built and maintained the entire backend infrastructure using Python and Sanic async microservices, managing device fleets, license authentication, and payment webhooks.
    • Created and owned Pokemod Atlas, building the Android application for automated device management and telemetry synchronization.
    • Worked on reverse engineering Unity IL2CPP game runtimes and runtime memory hooking with Frida.
  • HUNTME. Carrier onboarding and regulatory compliance platform for freight logistics.
    • Automated compliance verification for 7,500+ freight carriers through direct integration with the US FMCSA SAFER API.
    • Built real-time parser extracting safety scores, USDOT records, operating authorities, and active insurance policies.
    • Replaced manual compliance vetting workflows, eliminating 3 full-time operations roles.

Stack

  • Languages: Python, TypeScript, Kotlin, FunC
  • AI and voice: LiveKit, WebRTC, Gemini Live, Model Context Protocol (MCP), PyTorch, Qdrant, BGE-M3, RAG & Hybrid Search, Groq, Ollama, Hugging Face Transformers
  • Infrastructure: AWS (EKS, RDS, S3), Kubernetes, Terraform, Helm, Docker, PostgreSQL, Redis, Prometheus, Grafana, GitHub Actions, GitLab CI/CD
  • Frontend: Next.js, React, Tailwind CSS, HTML5 Canvas, WebGL, Three.js

Recognition

  • First place, Telegram Mini Apps Developer Contest (Callegram, 2023)
  • First place, Best Startup Pitching Competition (2024)
  • Honorable mention, Telegram Smart Contract Cracking Competition (2022)

Pinned Loading

  1. neuron-model neuron-model Public

    A Computational Substrate for Self-Organizing Biologically-Plausible AI

    Python 11 4

  2. active-inference active-inference Public

    Active inference sensorimotor simulations: artificial life built from PAULA neurons, biological connectomes (C. elegans), and MuJoCo physics

    HTML 5

  3. live-c-elegans live-c-elegans Public

    Real-time C. elegans active-inference simulation streamed over WebSocket to a browser canvas — add/remove food live

    TypeScript 2

  4. magic-prompt magic-prompt Public

    A tool that turns short, vague and context-less prompts into a structured, context-relevant and elaborate task descriptions for LLMs and Agents

    Python 6

  5. mcphero-cli mcphero-cli Public

    CLI tool for mcphero.app

    Python

  6. core_messenger core_messenger Public

    Open Messenger built on core_client

    Python 7