---
title: "Wesley Kambale (@kambale) | Hashnode"
description: "Machine Learning Engineer | AI Community Leader | Technical Educator. Read the latest articles by Wesley Kambale on Hashnode."
url: "https://hashnode.com/@kambale"
---

@kambale

# Wesley Kambale

@kambale · Mbarara, Uganda · Joined August 2021

Machine Learning Engineer | AI Community Leader | Technical Educator

[Website](https://kambale.dev/)

### [Never lose your progress: Checkpointing with Orbax](https://kambale.dev/orbax?utm_source=hashnode&utm_medium=feed)

Picture this. You have spent six hours training a model. The loss curve looks beautiful, accuracy is climbing, and you are one epoch away from a result worth writing home about. Then the power goes ou

16 min read · [kambale.dev](https://kambale.dev/orbax?utm_source=hashnode&utm_medium=feed)

[Aug 7](https://hashnode.com/posts/orbax/6a75ee4b5a7b977961af2731)

### [Feeding the Beast: Data Pipelines with Grain](https://kambale.dev/data-pipelines-with-grain?utm_source=hashnode&utm_medium=feed)

Over the last seven weeks we have built models, written training loops, composed optimizers, bulletproofed our code with Chex, and learned to debug inside JIT-compiled functions. Every one of those ar

13 min read · [kambale.dev](https://kambale.dev/data-pipelines-with-grain?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/data-pipelines-with-grain/6a6951a3c391c6185d3f46df) [Jul 29](https://hashnode.com/posts/data-pipelines-with-grain/6a6951a3c391c6185d3f46df)

### [When JIT hides your errors: Debugging JAX](https://kambale.dev/debugging-jax?utm_source=hashnode&utm_medium=feed)

Last week, Chex gave us a firewall against shape mismatches and NaN gradients. That firewall catches the bugs that announce themselves; a tensor with the wrong shape, a gradient that explodes into inf

13 min read · [kambale.dev](https://kambale.dev/debugging-jax?utm_source=hashnode&utm_medium=feed)

[Jul 18](https://hashnode.com/posts/debugging-jax/6a5ac4db01b9f03a874eeef9)

### [Catching bugs before they catch you](https://kambale.dev/chex?utm_source=hashnode&utm_medium=feed)

Here's a scenario every ML engineer eventually lives through: you kick off a training run before leaving the office. Three hours later, you check back in. The loss curve looks like a flat line at NaN.

14 min read · [kambale.dev](https://kambale.dev/chex?utm_source=hashnode&utm_medium=feed)

[Jul 12](https://hashnode.com/posts/chex/6a52e1a86ea286dd9ce4af68)

### [Optax: Optimizers You Can Compose Like LEGO](https://kambale.dev/optax-optimizers?utm_source=hashnode&utm_medium=feed)

In our last article on this blog in March, we built a complete training loop. We used optax.adamw() to create an optimizer, wrapped it in nnx.Optimizer, and watched our model learn. But we barely scra

11 min read · [kambale.dev](https://kambale.dev/optax-optimizers?utm_source=hashnode&utm_medium=feed)

[Jul 1](https://hashnode.com/posts/optax-optimizers/6a44cf56dcce9d94f5e9173e)
