Sandboxing an AI Agent
A guide to sandboxing AI agents: why an autonomous agent needs its own disposable computer, and the isolation tech underneath.
Sajal Sharma
Internet home of Sajal Sharma — AI engineer and O'Reilly instructor specializing in agentic AI systems, LLMs, and machine learning.
Latest posts
Agents Have Outgrown Workflows
Why the industry is shifting from handcrafted agentic workflows to agents operating inside harnesses, what that looks like in practice, and the trade-offs involved.
A Week with OpenClaw as My Personal Assistant
I spent the last week running my own personal AI assistant with OpenClaw. Here's what I built, what broke, and whether any of this is actually worth the effort and tokens.
2025: Career in Review
A reflection on a year of building AI products at a venture studio, teaching courses on O'Reilly, writing a viral blog post, and figuring things out.
Working Effectively with AI Coding Tools like Claude Code
A practical guide to working effectively with AI coding tools like Claude Code, covering mindset shifts, quality control strategies, and team collaboration workflows for modern software development.
Adventures with Claude Code: Reflections on Building a Full-Stack System with AI Assisted Coding
Thoughts after Claude Code for building a full-stack system, covering the productivity gains, challenges, and lessons learned from the frontier of AI-assisted software development.
Understanding MCP: How the Model Context Protocol Solves AI's Integration Problem
A complete overview of Model Context Protocol (MCP) and how it solves the M×N integration problem in AI development by creating a standardized interface between AI applications and external tools.
Agentic RAG Series - Part 3: Building a Comprehensive Agentic RAG Workflow: Query Routing, Document Grading, and Query Rewriting
A tutorial on building an advanced agentic RAG workflow that combines query routing, document grading, and query rewriting using LangGraph to create a robust, self-correcting retrieval system.
Agentic RAG Series - Part 2: Building an Agentic RAG Workflow with Query Router Using LangGraph
A coding tutorial on building an agentic RAG workflow with a query router using LangGraph, enabling the system to intelligently choose between Wikipedia and web search retrievers based on query type.
What's the Moat? Product Defensibility for AI Applications
Some thoughts on product defensibility for AI applications from my experience in the startup world.
Agentic RAG Series - Part 1: An Introduction to Agentic RAG
A comprehensive introduction to agentic RAG, common design patterns, as well as a few example pipelines.
Guest Lecture at Yale: February 2025 - Agentic Systems with LangGraph
Slides from my guest lecture at Yale for Generative AI & Entrepreneurship class (MGT 899).
Building AI Agents with LangGraph: My First O'Reilly Course!
Announcing my first video course with O'Reilly—Building AI Agents with LangGraph! This course dives deep into AI agent design, action-taking, and multi-agent architectures using Python and OpenAI.
An Overview of Multi Agent Frameworks: Autogen, CrewAI and LangGraph
A brief look at the components of multi-agent frameworks and the current cutting edge options.
Building a Corrective RAG workflow with LangGraph
A deep dive into the process building a corrective RAG workflow using langgraph to handle scenarios where the documents retrieved from a vector database in a traditional RAG workflow are not relevant to answer a question.
Deploy StableLM models on AWS Sagemaker Endpoints
This blog post guides you through the process of deploying StableLM models on AWS Sagemaker Endpoints, including creating a custom inference script and setting up the endpoint.
Building an Image Classifier Really Fast Using Fastai
In this post, I demonstrate how to quickly build an image classifier using the fastai library, a powerful tool for practical deep learning. The project involves classifying images of fruit as either rotten or fresh.
Coding K-Means Clustering using Python and NumPy
This post details the process of coding the K-Means Clustering algorithm from scratch using Python and NumPy. It's a great exercise for understanding the mechanics of this fundamental machine learning algorithm.