---
tags: Book
---

# Operator Playbook

## Scope of Work

- To maintain the continuous operation of any automated process
- To invoke services and direct their output within a pre-determined process

## Relevant Stories

- Making sure that the 'Retrieval Data' and the 'Base ML Process' execution flows runs without errors.
- This may require perform "quick fixes" if there's issues related to Ops. Else, the Subject Matter Expert and/or the Software / Data Engineers should be activated.
- Directly supporting the Human Evaluation Execution Flow.
- Directly supporting the Coordinator and the Subject Matter Expert to manually execute any independent microservice.

## Gitcoin SAD workflows to be executed

- At the beginning of the round
1. `prepare_prediction`
2. `prepare_training`
3. `fit`
4. `predict`
- Each time for preparing the human evaluations sheet
1. `prepare_prediction`
2. `predict`
3. `prepare_human_eval`
- After the human evaluations sheet has been populated.
1. `prepare_prediction`
2. `prepare_training`
3. `fit`
4. `predict`
- Each time that flags should be pushed to the API.
1. `prepare_prediction`
2. `predict`
4. `push_endpoint`