MLOps and Model Deployment Intern

By the end of six months

You will have taken three models that currently live in notebooks and put them behind a versioned, monitored serving stack that any engineer here can deploy without first finding a data scientist.

The stack and the work

The models are scikit-learn and PyTorch artefacts for job ranking, spam detection and salary estimation. You containerise them, wire up an experiment tracker so training runs stop getting lost, and build the retraining job that pulls fresh labelled data every week. Drift monitoring is yours too: you decide which input distributions are worth alerting on and set thresholds you can defend. Somewhere around month four you will break production, roll it back, and write the postmortem, which is a normal part of learning this properly.

Fit

This suits a final year student or recent graduate who enjoys plumbing as much as modelling. You need Python, comfort on a Linux shell, and enough Docker to be dangerous. You will pair with a senior platform engineer in the Hyderabad office two days a week and work remotely the rest, over a full six months. Nobody will hand you a sandbox copy of the system.