MPS Talk Series 2021. Day 3
summary
Modelix runs MPS headless on a server and exposes a ~500-line TypeScript client that captures input while the server implements all editor behavior, enabling real-time collaboration between browser and local MPS, branch-aware editing, per-user undo, and Git-like commits plus operation lists for convergence; models live on a central model server and scale via a Kubernetes cluster with delta images and partial loading. The Biological Knowledge Base uses a server-side projectional editor and a Kotlin DSL (btb language) with query dependency tracking, editor-service completions against large ontologies (NCBI, Cell Ontology, Reactome), OT-based sync, AWS deployment and planned text‑mining integration.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:11Model X goals
- 4:02Model X architecture
- 9:24Demo: Model X local
- 15:17Customizing web editors
- 24:31Debugging web editors
- 29:30Real-time collaboration
- 39:56Model X implementation
- 40:39Q&A with Sasha Lison
- 1:00:12Biological Knowledge Base project introduction
- 1:09:30Demo: Biological KB
- 1:24:10B2B language with Kotlin
- 1:42:13Projectional editing tech
- 1:57:33Q&A session with Biological Knowledge Base team
related talks
36mWhat's New, ModelMesh? Model Serving at Scale
33mBuilding Strimzi MCP Server for Kubernetes: Democratizing Platform Exper...
31mFrom CLI to MCP in 20 Minutes
34mOpen Source AI Agents on User-Owned Infra: K3s, MCP, & GPU Sharing in Practice
36mAccelerating High-Performance Machine Learning at Scale i...
28mExploring ML Model Serving with KServe (with fun drawings)
29mIntegrating High Performance Feature Stores with KServe Model Serving
23mProactive and Polymorphic Adaptation of Multi-Cloud Deployments