Home on Alex Kim's blog

The Claude Code Source Leak: fake tools, frustration regexes, undercover mode, and more

Tue, 31 Mar 2026 00:00:00 +0000

Update: see HN discussions about this post: https://news.ycombinator.com/item?id=47586778

I use Claude Code daily, so when Chaofan Shou noticed earlier today that Anthropic had shipped a .map file alongside their Claude Code npm package, one containing the full, readable source code of the CLI tool, I immediately wanted to look inside. The package has since been pulled, but not before the code was widely mirrored, including myself and picked apart on Hacker News.

This is Anthropic’s second accidental exposure in a week (the model spec leak was just days ago), and some people on Twitter are starting to wonder if someone inside is doing this on purpose. Probably not, but it’s a bad look either way. The timing is hard to ignore: just ten days ago, Anthropic sent legal threats to OpenCode, forcing them to remove built-in Claude authentication because third-party tools were using Claude Code’s internal APIs to access Opus at subscription rates instead of pay-per-token pricing. That whole saga makes some of the findings below more pointed.

How to Set Up Work and Personal Git Profiles

Fri, 25 Jul 2025 00:00:00 +0000

Tired of accidentally committing to work repos with your personal email? Here’s how to automatically use the right Git profile based on where your repositories live.

1. Generate SSH Keys

Create separate keys for each account:

# Work key
ssh-keygen -t ed25519 -C "work@company.com" -f ~/.ssh/id_ed25519_work

# Personal key
ssh-keygen -t ed25519 -C "personal@gmail.com" -f ~/.ssh/id_ed25519_personal

2. Configure SSH Hosts

Edit ~/.ssh/config:

# Work GitHub
Host github-work
HostName github.com
User git
IdentityFile ~/.ssh/id_ed25519_work
IdentitiesOnly yes
# Personal GitHub (default)
Host github.com
HostName github.com
User git
IdentityFile ~/.ssh/id_ed25519_personal
IdentitiesOnly yes

3. Set Up Git Profiles

Global config (~/.gitconfig):

US Tariffs, DeepSeek and OpenAI

Sun, 02 Feb 2025 00:00:00 +0000

This week the new Trump administration announced new tariffs on key trading partners, Canada and Mexico. Both countries vowed to retaliate, and tensions are high:

“Canada imposes 25% tariffs in trade war with US”

Living in Canada, I am obviusly curious about the impact this will all have both both sides of the border. I am a self-described economics nerd: having never studied economics, I try to read and listen about when I have spare time. One other topic that’s been on my mind recently is the recent release of the DeepSeek R1 model and the the release of the OpenAI O3 model that followed. Both are SOTA (as of Feb 2025) LLM models that were trained to “reason” before providing an answer through a chain of thought Reinforcement Learning:

Orchestrating LLM Fine-tuning on Kubernetes with SkyPilot and MLflow: A Complete Guide

Sat, 11 Jan 2025 00:00:00 +0000

Training and fine-tuning Large Language Models (LLMs) requires significant computational resources and careful experiment tracking. While many focus on the modeling aspects, efficiently managing compute resources and experiment tracking is equally important for successful ML projects. This guide demonstrates how to use SkyPilot and MLflow, two open-source tools, to orchestrate LLM fine-tuning jobs effectively.

An open-source stack for LLM fine-tuning

Modern LLM fine-tuning workflows involve multiple moving parts:

Resource orchestration across different cloud providers
Environment setup and dependency management
Experiment tracking and monitoring
Distributed training coordination
System metrics collection

Using SkyPilot for resource orchestration and MLflow for experiment tracking provides an easy-to-use and fully open-source stack for managing these complexities.

Kubernetes Mental Model

Sat, 04 Jan 2025 00:00:00 +0000

I am preparing for my CKAD (Certified Kubernetes Application Developer) exam. Below is the mental model of K8S concepts that helps me understand Kubernetes. Hope it helps you too.

The Big Picture: Kubernetes as an Orchestrator

What is Kubernetes?
Kubernetes is an automation system for deploying and managing containerized applications at scale. Rather than manually handling each container, you define your desired state—like “I want three replicas of my service running.” Kubernetes ensures this state remains true even if servers fail or traffic surges.

Intro to SLURM for ML Practitioners

Sun, 24 Nov 2024 00:00:00 +0000

SLURM (Simple Linux Utility for Resource Management) is an open-source workload manager designed to schedule and manage jobs on large clusters. In the world of LLMs, SLURM has seen a resurgence in popularity due to the increased demand for training large models and scaling them to multiple nodes.

This guide will introduce the fundamental concepts of SLURM, common commands and script structures, and show advanced scenarios like distributed multi-node training. I’ll also share some useful tips and tricks.

Experiments with OpenAI's Function Calling

Sun, 05 May 2024 00:00:00 +0000

Intro

This notebook (also on github) demonstrates how to use Function Calling functionality with the OpenAI API.

In this demo, we’ll use the Northwind database to convert natural language queries into SQL:

"What is the total revenue for each product in the database?" ->
-> "SELECT ... FROM ..." -> DataFrame

There will be two function calling examples:

A simple one-step function call to convert a natural language query into SQL, where we’ll put the database schema into the system prompt and them use function calling to convert a natural language query into SQL.
A two-step function call first gets the schema of the database and then converts a natural language query into SQL.

At the end, we’ll compare the two approaches and do a quick-and-dirty evaluation of the results using a hand-curated list of questions and their expected SQL queries in eval_questions.csv.

Fine-Tuning Large Language Models with a Production-Grade Pipeline

Fri, 08 Sep 2023 00:00:00 +0000

Introduction - Solving cloud resources and reproducibility for LLMs

A few of weeks ago, I wrote a post about the challenges of training large ML models, in particular:

the need for more computing power and the complexity of managing cloud resources;
the difficulty of keeping track of ML experiments and reproducing results.

There I proposed a solution to these problems by using SkyPilot and DVC to manage cloud resources and track experiments, respectively.

Trying to Understand Something Difficult? Minimize the Number of Attempts!

Wed, 16 Aug 2023 00:00:00 +0000

When you’re trying to wrap your head around challenging new ideas, your natural instinct may be to chip away at it in several attempts. Read a bit, take a break, come back later - like attacking a mountain ascent in short sprints. This may feel like you’re making progress, but you’re actually doing yourself a disservice. Every time you re-engage with the difficult material, you’re forcing your mind to re-enter a state of intense flow. This refocusing requires mental energy - energy that depletes your limited cognitive resources. Each time you dip back into flow, you have to:

ML experiments in the cloud with SkyPilot and DVC

Thu, 10 Aug 2023 00:00:00 +0000

Introduction

One of the things that makes machine learning hard is that you have to run a lot of experiments. You have to try different models, different data sets, different hyperparameters, different features. And each experiment can take a long time to run, especially if you’re working on deep learning problems. You can’t just run them on your laptop or desktop. You need more computing power, and you need it fast.

Why Sales Engineers Exist

Wed, 02 Aug 2023 00:00:00 +0000

Paul Graham once wrote that “a startup is a company designed to grow fast.” For a startup to grow fast, especially a B2B startup selling SaaS products to other companies, it needs sales engineers.

Sales engineers are the technical experts who work closely with sales teams. They exist because selling enterprise software requires deep technical knowledge that salespeople typical don’t have. The salesperson establishes rapport with the prospect and understands their business needs. But when it comes time to demo the product and answer nitty gritty technical questions, that’s where the sales engineer comes in.

Don’t know what to do next? Teach!

Mon, 31 Jul 2023 00:00:00 +0000

We (esp. those in the tech industry) have an instinctual aversion to pedagogy. “Those who can’t do, teach” the old saying goes. But the truth is the opposite. Teaching does not indicate an inability to do something. On the contrary, teaching empowers and enables ability. It is the highest form of understanding. When you teach something, you gain a deeper mastery over the subject matter than you would as a passive student. As Leonardo da Vinci said:

About

Mon, 01 Jan 0001 00:00:00 +0000

Dad | ML Engineer | Technical Instructor | Consultant | Community Builder

Hi! I’m Alex.

I am an ML engineer by trade and a physicist by degree. I’ve lived in different parts of the world and am currently based in Montreal, 🇨🇦. I speak English, Russian, French and Spanish. I am also a husband and a dad of two 👧🏻👦🏻.

Most of my recent work sits at the intersection of MLOps and generative AI - building distributed training infrastructure for large models and deploying LLM-powered systems in production.

Week 1: Kick-starting an ML project

Mon, 01 Jan 0001 00:00:00 +0000

Slides 🖼️

Week 1: ML project lifecycle and MLOps best practices

Learning objectives

Understand the core philosophy behind MLOps ideas
Apply best practices for establishing ML project structure and dependencies management
Manage project dependencies with pip and virtualenv
Version datasets with DVC

Project Introduction

Problem Description and Dataset

This dataset contains 10,000 records, each of which corresponds to a different bank’s user. The target is Exited, a binary variable that describes whether the user decided to leave the bank. There are row and customer identifiers, four columns describing personal information about the user (surname, location, gender, and age), and some other columns containing information related to the loan (such as credit score, the current balance in the user’s account and whether they are an active member among others).

Week 2: ML Pipelines, Reproducibility Experimentation

Mon, 01 Jan 0001 00:00:00 +0000

Slides 🖼️

Week 2: ML Pipelines, Reproducibility and Experimentation

Learning objectives

Refactor a Jupyter notebook into a reproducible ML pipeline
Version artifacts of an ML pipeline in a remote storage
Iterate over a large number of ML experiments in a disciplined way

Steps

Refactor Jupyter notebook in a DVC pipeline

Docs: https://dvc.org/doc/start/data-pipelines

Create the following files to read parameter values from a file

params.yaml

 base:
project: bank_customer_churn
raw_data_dir: data/raw
countries:
- France
- Spain
feat_cols:
- CreditScore
- Age
- Tenure
- Balance
- NumOfProducts
- HasCrCard
- IsActiveMember
- EstimatedSalary
targ_col: Exited
random_state: 42
data_split:
test_size: 0.25
processed_data_dir: data/processed
train:
model_type: randomforest
model_dir: models
model_path: models/clf-model.joblib
params:
n_estimators: 200
max_depth: 20

Week 3: CI/CD for ML and ML-based Web API

Mon, 01 Jan 0001 00:00:00 +0000

Slides 🖼️

Week 3: CI/CD for ML

Learning Objectives

Learn the basics of CI/CD
Leverage the power of CI/CD tools for ML projects with CML
Integrate an ML model into the FastAPI framework
Build and test a Docker container running a web API service
Deploy the resulting Docker container to cloud

Steps

Introduction to GitHub Actions and CML

Introduction to GitHub Actions
Introduction to CML

Add PERSONAL_ACCESS_TOKEN , AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY to GH secrets: https://docs.github.com/en/actions/security-guides/encrypted-secrets
- For AWS credentials, see https://docs.aws.amazon.com/sdk-for-javascript/v2/developer-guide/getting-your-credentials.html
- For PERSONAL_ACCESS_TOKEN:
  - Generate a new personal access token under GitHub developer settings
  - in the “Note” field, type PERSONAL_ACCESS_TOKEN
  - select repo scope
  - click “Generate token” and copy it
  - In your GitHub repository and/or organization, navigate to Settings -> Secrets -> New repository/organization secret
  - in the “Name” field, type PERSONAL_ACCESS_TOKEN
  - in the “Value” field, paste the token
  - click Add secret

Create.github/workflows/train-model.yaml

name: train-model
on:
push:
paths:
- "data/**"
- "src/**"
- "params.yaml"
- "dvc.*"
jobs:
train-model:
runs-on: ubuntu-latest
environment: cloud
permissions:
contents: read
id-token: write
steps:
- uses: actions/checkout@v3
with:
ref: ${{ github.event.pull_request.head.sha }}
- uses: iterative/setup-cml@v1
- uses: actions/setup-python@v2
with:
python-version: "3.10"
- uses: actions/setup-node@v1
with:
node-version: '16'
- name: SetupGitUser
run: cml ci
env:
REPO_TOKEN: ${{ secrets.PERSONAL_ACCESS_TOKEN }}
- name: TrainModel
env:
REPO_TOKEN: ${{ secrets.PERSONAL_ACCESS_TOKEN }}
AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
run: |
pip install -r requirements.txt
dvc pull
dvc exp run
dvc push
# Create CML report
echo "## Metrics" >> report.md
dvc metrics show --md >> report.md
echo "## Feature Importances" >> report.md
csv2md reports/feat_imp.csv >> report.md
echo "## Confusion Matrix" >> report.md
echo '![](reports/figures/cm.png)' >> report.md
cml comment create report.md

Push workflow file with git
Modify some model parameters (e.g. max_depth), rerun the pipeline (dvc exp run) and push changes to DVC remote and git
Review GitHub Actions runs

Web App Development

Create web application src/app/main.py

import json
import sys
from pathlib import Path
import uvicorn
src_path = Path(__file__).parent.parent.resolve()
sys.path.append(str(src_path))
from typing import List
import pandas as pd
from fastapi import Body, FastAPI, Request
from fastapi.middleware.cors import CORSMiddleware
from joblib import load
from pydantic import BaseModel
from utils.load_params import load_params
app = FastAPI()
# https://fastapi.tiangolo.com/tutorial/cors/#use-corsmiddleware
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
params = load_params(params_path='params.yaml')
model_path = params.train.model_path
feat_cols = params.base.feat_cols
model = load(filename=model_path)
class Customer(BaseModel):
CreditScore: int
Age: int
Tenure: int
Balance: float
NumOfProducts: int
HasCrCard: int
IsActiveMember: int
EstimatedSalary: float
class Request(BaseModel):
data: List[Customer]
@app.post("/predict")
async def predict(info: Request = Body(..., example={
"data": [
{
"CreditScore": 619,
"Age": 42,
"Tenure": 2,
"Balance": 0,
"NumOfProducts": 1,
"HasCrCard": 1,
"IsActiveMember": 1,
"EstimatedSalary": 101348.88
},
{
"CreditScore": 699,
"Age": 39,
"Tenure": 21,
"Balance": 0,
"NumOfProducts": 2,
"HasCrCard": 0,
"IsActiveMember": 0,
"EstimatedSalary": 93826.63
}
]
})):
json_list = json.loads(info.json())
data = json_list['data']
input_data = pd.DataFrame(data)
probs = model.predict_proba(input_data)[:,0]
probs = probs.tolist()
return probs
if __name__ == "__main__":
uvicorn.run(app, host="0.0.0.0", port=8000)

Test API

Week 4: Monitoring for ML Projects

Mon, 01 Jan 0001 00:00:00 +0000

Slides 🖼️

Week 4: Data Drift Monitoring for ML Projects

Learning Objectives

Distinguish between application monitoring and ML monitoring
Use Alibi Detect framework to detect data drift

Steps

Introduction to Data Drift Monitoring

What’s data drift and why do we need to monitor for it?
Intro to Alibi Detect
Add Churn_Modelling_Germany.csv to data/more_data/

Churn_Modelling_Germany.csv
Add /more_data entry to data/.gitignore
Create and explore notebooks/DriftDetection.ipynb

DriftDetection.ipynb

Incorporate drift detection into the DVC pipeline

Create src/stages/drift_detector.py

import sys
from pathlib import Path
src_path = Path(__file__).parent.parent.resolve()
sys.path.append(str(src_path))
import argparse
import pandas as pd
from alibi_detect.cd import TabularDrift
from alibi_detect.saving import save_detector
from joblib import load
from utils.load_params import load_params
def train_drift_detector(params):
processed_data_dir = Path(params.data_split.processed_data_dir)
model_dir = Path(params.train.model_dir)
model_path = Path(params.train.model_path)
model = load(model_path)
X_test = pd.read_pickle(processed_data_dir/'X_test.pkl')
X_train = pd.read_pickle(processed_data_dir/'X_train.pkl')
X = pd.concat([X_test, X_train])
feat_names = X.columns.tolist()
preprocessor = model[:-1]
categories_per_feature = {i:None for i,k in enumerate(feat_names) if k.startswith('cat__')}
cd = TabularDrift(X,
p_val=.05,
preprocess_fn=preprocessor.transform,
categories_per_feature=categories_per_feature)
detector_path = model_dir/'drift_detector'
save_detector(cd, detector_path)
if __name__ == '__main__':
args_parser = argparse.ArgumentParser()
args_parser.add_argument('--config', dest='config', required=True)
args = args_parser.parse_args()
params = load_params(params_path=args.config)
train_drift_detector(params)

Add train_drift_detector stage to dvc.yaml

Work With Me

Mon, 01 Jan 0001 00:00:00 +0000

I am available for consulting and training engagements.

Organizations I’ve worked with include: McKinsey & Company, Concordia Bootcamps, Splunk, BloomTech, TripleTen, O’Reilly Media, Iterative.ai, Platacard.mx, Nebius, Grainger and others.

Client testimonials

Anton Tarasenko, Co-founder and CBPO of Platacard.mx:

Home on Alex Kim's blog

The Claude Code Source Leak: fake tools, frustration regexes, undercover mode, and more

How to Set Up Work and Personal Git Profiles

1. Generate SSH Keys

2. Configure SSH Hosts

3. Set Up Git Profiles

US Tariffs, DeepSeek and OpenAI

Orchestrating LLM Fine-tuning on Kubernetes with SkyPilot and MLflow: A Complete Guide

An open-source stack for LLM fine-tuning

Kubernetes Mental Model

The Big Picture: Kubernetes as an Orchestrator

Intro to SLURM for ML Practitioners

Experiments with OpenAI's Function Calling

Intro

Fine-Tuning Large Language Models with a Production-Grade Pipeline

Introduction - Solving cloud resources and reproducibility for LLMs

Trying to Understand Something Difficult? Minimize the Number of Attempts!

ML experiments in the cloud with SkyPilot and DVC

Introduction

Why Sales Engineers Exist

Don’t know what to do next? Teach!

About

Week 1: Kick-starting an ML project

Slides 🖼️

Learning objectives

Project Introduction

Problem Description and Dataset

Week 2: ML Pipelines, Reproducibility Experimentation

Slides 🖼️

Learning objectives

Steps

Refactor Jupyter notebook in a DVC pipeline

Week 3: CI/CD for ML and ML-based Web API

Slides 🖼️

Learning Objectives

Steps

Introduction to GitHub Actions and CML

CI/CD: Automatic reporting for model-related changes

Web App Development

Week 4: Monitoring for ML Projects

Slides 🖼️

Learning Objectives

Steps

Introduction to Data Drift Monitoring

Incorporate drift detection into the DVC pipeline

Work With Me