Technical Homework Senior Data Scientist

Deployment FastAPI via Heroku

📝 Table of Contents

• About the Optims
• About the Deployment
• Geting Started

👨‍🔬 About the Optims

All the logic falls under the app folder. To the date, we have develope 3 optimizatión objects. OptimExp, OptimNegBinomial and OptimStochConstraint

They make use of Bayesian optimization, distribution boostraping and linear optimization to address the problem of a cold start agent, without data or defined environment to learn from. We should model not only the number of new candidates impacted each call from the pool, but the callback in minutes till the next one.

All of them have the invitation_logic_api() method, which receive the same input:

Field	Description	Type	Example
correlation_id	Event id. Case{n case}_{n call}	string	"Case0_2"
reference_date_time	api call timestamp	string (timestamp YYYYMMDD HH:MM:SS)	"2021-01-02 03:12:27"
deadline	job opening deadline timestamp	string (timestamp YYYYMMDD HH:MM:SS)	"2021-01-05 09:00:00"
num_vacancies	job opening number of vacancies (constant)	int	8
num_remaining_in_pool	remaining pool workers after the invitations already sent	int	98
impacted_candidates_data	information about the impacted candidates. A list of dictionaries containing their current status and how long it took to receive the reply (so far)	list	[{"notification_status": "ir_rejected", "candidate_status": "not_in_ft", "time_to_respond_ir_minutes": 1 }]

🧐 About the Deployment

We have make available the optim engines throuhg API endpoints deployed in Heroku accessible URI:

https://jandt-technical-carlosvecina.herokuapp.com/

You could find API documentation:

https://jandt-technical-carlosvecina.herokuapp.com/docs https://jandt-technical-carlosvecina.herokuapp.com/redoc

Each repo is the source to the automatic integration FastAPI -> Heroku. The app automatically deploys from this repo main branch.

You could access to the API optim-stoch-constraint endpoint in Python using:

import requests
from datetime import datetime as dt

optim = 'optim-nbinomial'
url = f'https://jandt-technical-carlosvecina.herokuapp.com/{optim}/'

data = {'now':str(datetime.datetime(2021,11,1)),
        'deadline': str(datetime.datetime(2021,11,2)),
        'num_vacancies': 10,
        'num_remaining_in_pool':500,
        'impacted_candidates_data': [        {
            "notification_status": "ir_accepted",
            "candidate_status": "offer_accepted",
            "time_to_respond_ir_minutes": 48
        },
        {
            "notification_status": "ir_rejected",
            "candidate_status": "not_in_ft",
            "time_to_respond_ir_minutes": 11
        }]
       }

resp = requests.post(url, json=data)
print(resp.json())

🏁 Getting Started

These instructions will get you a copy of the project up and running on your local machine.

Installing

Requirements

• Docker >= 17.05
• Python >= 3.7
• Poetry

NOTE - Run all commands from the project root

Local development

Poetry

Create the virtual environment and install dependencies with:

poetry install

See the poetry docs for information on how to add/update dependencies.

Spawn a shell inside the virtual environment with:

poetry shell

Start a development server locally:

poetry run uvicorn app.main:app --reload --host localhost --port 8000

API will be available at localhost:8000/

Docker

Build images with:

docker build --tag poetry-project --file docker/Dockerfile .

The Dockerfile uses multi-stage builds to run lint and test stages before building the production stage. If linting or testing fails the build will fail.

You can stop the build at specific stages with the --target option:

docker build --name poetry-project --file docker/Dockerfile . --target <stage>

For example we wanted to stop at the test stage:

docker build --tag poetry-project --file docker/Dockerfile --target test .

We could then get a shell inside the container with:

docker run -it poetry-project:latest bash

Credits

• Poetry Docker Heroku buildpack Michael Oliver