---
title: "PyTorch | Skills We Assign For | Azendo"
description: "PyTorch for model work that reaches production — training, serving and the Azendo roles assigned for AI engineering."
url: "https://azendo.co/skills/pytorch/"
---

[Skills](https://azendo.co/skills/) AI and machine learning 

# PyTorch.

PyTorch is a deep learning framework used to define, train and run neural networks. It dominates research and has become the default for production work involving language models and computer vision.

## Where PyTorch fits on a long engagement.

Most commercial PyTorch work is not training from scratch. It is fine-tuning an existing model, adapting one to a specific domain, or getting something that works in a notebook to run reliably under load. The interesting problems are usually inference cost and latency, not architecture.

The gap between a training script and a served model is where projects stall. Batching, memory, versioning and the difference between how the model behaved on the evaluation set and how it behaves on real inputs are all engineering problems rather than modelling ones.

## What an assigned team does with PyTorch.

Model work has a long tail that project budgets never cover. The demo takes weeks, the thing that survives real inputs takes months, and the difference is almost entirely evaluation and inference engineering rather than modelling.

That tail is why this capacity suits a standing assignment. Where the models are already live and the problem is keeping them working, the assignment is [MLOps engineering](https://azendo.co/services/mlops-engineering/) rather than model development.

## What we use PyTorch for.

* Fine-tuning on domain data Adapting a general model to a specific vocabulary or document type, where an off-the-shelf model is close but not close enough.
* Getting inference cost down Quantisation, batching and hardware choices that make a working model affordable to run continuously.
* From notebook to service Model code restructured so it can be versioned, tested and deployed like the rest of the system.

## How PyTorch capacity is assigned.

Model work is assigned under [AI engineering services](https://azendo.co/services/ai-engineering/), and where the constraint is keeping models running rather than building them, under MLOps engineering.

## Roles we assign PyTorch for

* [Computer Vision Engineer AI engineering](https://azendo.co/services/ai-engineering/computer-vision-engineer/)
* [LLM Engineer AI engineering](https://azendo.co/services/ai-engineering/llm-engineer/)
* [Machine Learning Engineer MLOps engineering](https://azendo.co/services/mlops-engineering/machine-learning-engineer/)

## Service lines it sits in

* [AI engineering](https://azendo.co/services/ai-engineering/)
* [MLOps engineering](https://azendo.co/services/mlops-engineering/)

Capacity is agreed as a committed monthly capacity across a discipline, not per skill.

## Related in ai and machine learning

* [TensorFlow — skill we assign for](https://azendo.co/skills/tensorflow/)
* [scikit-learn — skill we assign for](https://azendo.co/skills/scikit-learn/)
* [Pandas — skill we assign for](https://azendo.co/skills/pandas/)
* [NumPy — skill we assign for](https://azendo.co/skills/numpy/)
* [embeddings — skill we assign for](https://azendo.co/skills/embeddings/)
* [function calling — skill we assign for](https://azendo.co/skills/function-calling/)
* [vLLM — skill we assign for](https://azendo.co/skills/vllm/)
* [MLflow — skill we assign for](https://azendo.co/skills/mlflow/)

## Tell us what your roadmap needs PyTorch for.

A service delivery manager replies with the disciplines we would assign, the monthly capacity and what the first month looks like.

[All skills we assign for](https://azendo.co/skills/)