Skip to main content

Welcome!

Welcome to Synthetic! :) Hopefully these docs will help make your onboarding experience smoother and more fun.

These Docs​

In the philosophy of docs-as-code, these docs live under docs in our main repo. PRs that change code should also include updates to any relevant docs.

tip

Docs are a powerful resource but inevitably get outdated. Please feel empowered to take ownership and update these docs to keep them useful, or improve them with anything others may find helpful!

Context​

I (Billy) always find it useful to have some historical and social context when joining a new team.

Synthetic Lab was originally started by Matt, who was experimenting with LLMs and open-source LLMs for an LLM-powered tabletop-like dungeon DM game he was building.

Simultaneously, Billy was working on an LLM-powered NSFW stable-diffusion based game.

We both found that while open-source LLMs had a lot of potential and features some closed source frontier models did not, they were hard and complex to run and manage, even for "technical" users.

Thus, Synthetic (formerly glhf.chat) was born! Matt started working on the idea around May of 2024, and Billy joined a couple months later, and the rest is history. :)

Matt is an ex-Instgram and ex-Airbnb early engineer, and Billy is an ex-Google eng.

Services​

Depending on your role, you may need access to different third-party services. A service admin (probably Matt or Billy) should create accounts for you or invite you to the necessary ones.

This list is as much to remind us us as it is for you. :P

Please ping one of us if you think you need access to one of these or we've forgotten to add you to any. :)

General​

Eng - General​

Eng - Production Cloud​

Eng - Production Services​

  • Clerk - Authentication and user management
  • Stripe - Payment services
  • Loops - Email newsletters for end-users

Eng - Observability​

  • Mixpanel - Event-based metrics
  • Honeycomb - OpenTelementry metrics and alerts
  • Grafana - Observability for Kubernetes and other production dashboards.
  • Sentry - Application error tracking

Eng - DevOps​

Eng - LLM Inference/GPU hosting​

  • Together.ai - Inference provider

    • We also use Together AI Managed Kubernetes via contract and direct POC (dedicated GPU hosting)
    • SSH and Kubernetes configs 1Password.

  • Fireworks.ai - Inference provider

  • RunPod - On-demand GPUs

  • Nebius - On-demand GPU cloud we use for self-hosting some models.

  • Fly.io

  • Lambda Labs

Marketing​

  • Syften - Live social media keyword monitoring.
  • Loops - Email campaigns (welcome, newsletter)

Company services​

Eng​

These are some helpful resources for technologies we use often, but feel free to dive straight into the code!

Getting Started​

You probably know better than this document on what you need to get started, but here are some suggestions in no particular order:

  • Say hello in #general on Slack!
  • Take a look at Getting Started and try to get our app running!
  • Submit your first PR!