We use analytics cookies.We use cookies to analyze site traffic and improve your experience.
Loading...
System Design and AI Engineering, by Roni Das. 984 interactive lessons.
Start LearningDesign the systems behind the apps you use every day
770 interactive lessons, from your first load balancer to Staff-level trade-offs. Every diagram is drawn step by step, like the one on this page. Press the button and grow an app yourself.
one time · $25 worldwide · ₹1,111 in India · no sign-up to start
step 6 · a CDN
What broke: Users far away wait for every image and file.
What we add: A CDN keeps copies of files in cities near the users.
Taught in: Web & Content Delivery · 31 lessons
the syllabus
Each rack is one level. Each row is a topic, and each bar is 10 lessons. The first 18 lessons are free, and the last 20 are whole designs, from a URL shortener to a payment system.
119 lessons · 4 of 23 topics
449 lessons · 12 of 23 topics
182 lessons · 6 of 23 topics
20 whole designs
Not videos. Not a wall of text. You watch real architectures get built one component at a time, with the production trade offs, the failure modes, and the diagrams drawn the way a senior engineer draws them on a whiteboard.
example · a video streaming service
traffic flowing through it · every box is a lesson
This is how every concept is taught inside the course. Not a static slide, a moving diagram you can follow.
One front door, many servers. Spread traffic so no single box ever falls over.
Answer from memory in microseconds. Touch the database only when you truly must.
Split one giant table across machines by key, so it scales far past one server.
Serve every user from the location closest to them, anywhere in the world.
Decouple producers from consumers so a traffic spike never takes you down.
Reason about throughput and tail latency, not just boxes and arrows.
The first lessons assume you have never designed a backend. The last ones get into the consensus and consistency problems that trip up Staff candidates. You start where you are, and the path is already laid out.
The same infrastructure you will learn to reason about
The gap between memorizing an architecture and designing one under pressure is reps. Every lesson makes you do the work.
You step through every diagram as it is drawn, run the code, and answer a quiz that proves you understood it. Reading a definition is not the same as being able to design the thing.
Full teardowns of real outages, like the Cloudflare November 2025 incident, so you learn how systems break under load and bad config, not just how they look on a clean diagram.
770 lessons across 23 topics, in order. Latency and caching at the start, consistency models and consensus by the end. No prior distributed systems background assumed.
Every pattern an interviewer actually probes: shard keys, cache invalidation, message queues, rate limiting, replication lag, and the trade offs behind each call.
Reviews
Reviews from readers of the courses, shown exactly as they wrote them. Each one shows how many lessons that person had read when they wrote it, counted from their own reading.
5.0 from 5 reviews
It is easy and good
It is really nice and good Please join the course
The concepts were explained in a clear and structured way, with practical examples that made even complex system design topics easier to understand. I especially liked the focus on real-world architecture, scalability, trade-offs. The content was well designed, engaging, and highly useful for anyone looking to strengthen their system design skills. Highly recommended for software engineers preparing for system design interviews or wanting to build a stronger foundation in designing scalable systems.
Outstanding all-in-one source of knowledge, especially excellent practice
Ive started this course by accident - LLM advised me this site. When I passed few free lessons, I had no doubt - this course should be bought. And I did it with special pleasent price and it was best decision of the month. Content is great, interesting to read and with a lot of practial cases (what makes it different from other souces and books) I hope to teach all the lessons, that will be my supergoal. Glad that I found this resource and speacial Thank You to Roni Das - Great Job!
The material has been designed with a lot of effort and dedication. It has covered vast range of topic with sufficient content and diagrams wherever required. It complements well with other system design materials.
Every lesson, every future lesson, no subscription. Try the first 18 free, then unlock the rest.
pay by UPI, card or netbanking in India · pay by card or PayPal everywhere else
course 1
From absolute beginner to principal engineer, drawn step by step.
System design is the skill of deciding how a software system is built so it keeps working as more people use it: which parts it has, how they talk to each other, where the data lives, and what happens when one part fails. It is what a system design interview tests, and it is what senior engineers do every day at work.
This course teaches it in order, in 770 lessons across 23 topics. You start with latency, load balancers and caches, move to databases, messaging, APIs and microservices, and end with consensus, consistency models and reliability. Then you put it together in whole designs, from a URL shortener to a payment system. Each lesson draws its diagram step by step, explains every term the first time it appears, and ends with a quiz.
One course, four ways in. Find the line that sounds like you, and it tells you where to start and what to spend your time on.
Start at lesson 1 of the foundation level (119 lessons, about 48 hours).
You need to be able to write a small program in any language. You do not need to have built a backend or used a cloud. The first lesson is latency, and it starts with what a tenth of a second of delay did to Amazon’s sales before it defines anything. Every term is explained the first time it is used. The level has 4 chapters (lesson counts in brackets): Core Fundamentals (26), Load Balancing & Proxies (26), Caching Strategies (36), Database Fundamentals (31).
Foundation quickly, then databases, messaging and APIs, then the capstone designs.
Most designs an interviewer asks for, from a URL shortener to a chat app, are built from the same parts: Database Types & Storage (39), Data Replication & Distribution (31), Messaging & Event Systems (41), API Design & Protocols (28). Learn those, then do the 20 capstone designs out loud before you read each answer. If your round is a machine coding or low level design round (classes and patterns, not servers), that is a different interview: see our low level design questions. For the mid-level round specifically, read the SDE-2 system design guide.
Skim what you know, then spend your time in the advanced level (182 lessons, about 88 hours).
Senior rounds go deeper than boxes and arrows: what happens to a write when a replica is behind, how a leader is elected, what exactly-once really costs, how a system degrades instead of falling over. That is Distributed Systems Core (24), Consistency Models (21), Stream & Batch Processing (36), Reliability & Resilience (85).
Read in order, and slow down in the intermediate level (449 lessons, about 131 hours).
Interview prep usually stops at the design. Production work starts there: Cloud Infrastructure (45), Security Architecture (57), Observability & Monitoring (60), DevOps & CI/CD (36). These are the chapters to read slowly, next to the systems your team runs.
Here is one real lesson, taken apart: Cache Eviction Policies, from the caching chapter. It is one of the free lessons, so you can open it and check everything below.
A cache with a size limit has to answer one question over and over: when it is full and something new arrives, what goes?
What the lesson is really about, in its own summary: LFU beat LRU by 6 to 15 points, FIFO measured identical to random, and the scan that supposedly destroys LRU cost it 11 points for four hundred requests. The numbers come from a small program in the lesson that replays 200,000 requests through each policy, not from a textbook, and the lesson shows the run.
The figures carry the explanation. This lesson uses 8 kinds: side-by-side comparisons, a recorded lab run, charts of measured results, hand-drawn sketches, annotated summaries, sequence diagrams, architecture diagrams with real product logos, flowcharts. Each figure has a short note under it saying what to notice, like this one:
Every one of the course’s lessons ends with a short quiz, and each answer explains why. This is the first of the 4 questions in this one:
On Zipf-distributed traffic with a cache holding 50 of 5,000 keys, how did LFU compare to LRU?
Why: A Zipf head is defined by frequency, which is what LFU measures, so it holds those keys permanently. LRU evicts a hot key as soon as enough other keys are touched after it, and in a small cache that happens quickly. The gap narrows as the cache grows: 6.2 points at 1,000 entries.
Beside every lesson there is an AI tutor that answers from the lesson you are reading, so a question like “why not just make the cache bigger?” is answered in the context of this lesson. Not every lesson is built like this one: some are shorter, some have code you can run in the browser, and the capstones are whole designs from requirements to trade-offs.
These 18 lessons are free to read in full, with no signup and no card: about 7 hours of reading across Core Fundamentals, Load Balancing & Proxies and Caching Strategies. They are the real lessons, not cut-down previews, so they are the honest way to decide. Every other lesson in the course shows its opening part free, before the paywall.
Sequential request-response pattern where each operation waits for the previous one
Non-blocking concurrent operations that don't wait for each other
The internet's phone book, how domain names become IP addresses
The fastest cache you can have, and the only one that can disagree with itself
The cache your own code manages, the race hiding in its default pattern, and how often that race actually fires
The only cache with no purge, the only one that can answer with no network, and why revalidation gives up the thing that makes it valuable
A cache your code keeps on the user's device, why the usual objection to it is measurably wrong, and what actually limits it
How to choose a cache TTL from your own traffic. A sweep of seven TTL values on skewed traffic shows the hit rate ceiling and the cost of staleness.
The one cache that cannot serve a wrong answer, what its lookup actually costs, and why its table is never freed
A shared cache is a network service, so how you call it matters far more than which product you picked
Caching the finished response removes all the work, and one personalised element removes all the benefit
Your database already cached the disk, so what a query cache saves is CPU, and that varies by 430 times between queries
Caching a result set freezes it, and the deep pagination it is usually bought to fix disappears when you change the question instead
LFU beat LRU by 6 to 15 points, FIFO measured identical to random, and the scan that supposedly destroys LRU cost it 11 points for four hundred requests
Careful per-row invalidation served more than twice the stale reads of doing nothing, because it raised the hit rate on the entries it cannot see
If they work for you, the whole course is ₹1,111 once in India or $25 once everywhere else, with lifetime access and every lesson added later.
The real chapters, in the order the course teaches them. The first lessons of each are listed, and the rest of every chapter opens below it, so this is the whole course, lesson by lesson. Lessons marked free can be read now.
The building blocks. Start here if you have never designed a backend.
Open them from the Core Fundamentals chapter.
Open them from the Load Balancing & Proxies chapter.
Open them from the Caching Strategies chapter.
Open them from the Database Fundamentals chapter.
Databases, messaging, APIs, microservices, cloud, security and observability: the parts every real system is made of.
Open them from the Database Types & Storage chapter.
Open them from the Data Replication & Distribution chapter.
Open them from the Messaging & Event Systems chapter.
Open them from the API Design & Protocols chapter.
Open them from the Microservices Architecture chapter.
Open them from the Data Governance & Compliance chapter.
Open them from the Cloud Infrastructure chapter.
Open them from the Kubernetes & Containers chapter.
Open them from the Security Architecture chapter.
Open them from the Observability & Monitoring chapter.
Open them from the DevOps & CI/CD chapter.
Open them from the Web & Content Delivery chapter.
Distributed systems, consistency, stream processing and reliability: the questions senior and staff interviews go deep on.
Open them from the Distributed Systems Core chapter.
Open them from the Consistency Models chapter.
Open them from the Stream & Batch Processing chapter.
Open them from the Security Testing & Operations chapter.
Open them from the Reliability & Resilience chapter.
Whole systems designed from requirements to trade-offs, the way an interview asks for them.
If your interview is a few months away, read the foundation level in order, then the parts of the intermediate level your target role uses most (databases, messaging and APIs come up in almost every design), then the capstone designs. Practise each capstone out loud before you read our answer. The free system design interview questions page and the interview walkthroughs are good for the last weeks.
If you are learning for your job, follow the course in order and take the chapters your team depends on more slowly: observability, security, DevOps and cloud infrastructure are where interview prep usually stops and production work begins. The course is about 287 hours at the estimates on each lesson, and you keep it for life, so there is no rush.
Comparing options first? Our best system design courses page puts 9 of them side by side, with sources, and says plainly where another one may suit you better.
₹1,111 once in India or $25 elsewhere. It is one payment, not a subscription, and it gives lifetime access to all 770 lessons, including every lesson added later. 18 lessons are free to read first with no signup.
18 lessons of this course are free with no signup, and so are our 106 interview walkthroughs at /interview and the system design interview questions page. That is enough to learn the basics and see how the course teaches. The full course of 770 lessons is paid.
It depends on how you learn, your budget and how soon your interview is. We compared 9 options, including ours, with each price checked on the provider's own page, on our best system design courses page. We make this course, so read that comparison with our bias in mind.
Learn the building blocks first: latency, load balancing, caching and databases. Then learn how the pieces talk to each other: queues, APIs, replication and sharding. Then practise whole designs, out loud, from requirements to trade-offs. This course is laid out in that order, and every lesson ends with a quiz so you know it stuck.
It feels hard because there is no single right answer, and the topics are wide. It gets easier when you learn it in order and draw each system as you go. The first lessons here assume you have never designed a backend, and every term is explained before it is used.
You should be able to write a small program in any language and know roughly what a web request and a database table are. You do not need to have designed a backend, used a cloud provider or worked at a large company. The foundation level starts from latency and builds up from there.
The whole course is about 287 hours at the estimates on each lesson, so nobody finishes it in a few days. For an interview, the 119 foundation lessons and the 20 capstone designs are the core; many engineers spend a few months of steady practice before they feel ready. Being good at it at work takes longer, because you learn the most from real systems.
Four levels: Foundation (core fundamentals, load balancing, caching, databases), Intermediate (database types, replication, messaging, APIs, microservices, cloud, Kubernetes, security, observability, DevOps and content delivery), Advanced (distributed systems, consistency models, stream and batch processing, reliability) and 20 capstone designs. The full list is on this page.
Many companies include a system design round when they hire experienced software engineers, and the same skills decide how well real systems hold up at work. It matters more as you move from junior to senior roles, because you are expected to make the design decisions, not only write the code.
The self-paced course does not give a certificate. The live System Design Interview Bootcamp does give a certificate of completion at the end of the programme. It is a separate live programme with its own fee (see /live), starting on Sunday, 8 November 2026, with 14 live classes (Sundays, one a week) and two 1:1 mock interviews, and it includes this course.
Yes. It is fully online. In India you pay ₹1,111 once in rupees, by UPI, card or netbanking. Everywhere else it is $25 by card or PayPal.
No. Lessons are written and drawn: each diagram builds step by step as you read, some lessons have code you can run, and every lesson has a quiz. An AI tutor sits beside every lesson and answers from the lesson you are reading.
Yes. Both courses together cost ₹1,999 in India or $49 elsewhere, in one payment with lifetime access to both. Bought one by one they cost ₹2,444 in India or $60 elsewhere.
NEWSystem Design Hands-On
Postgres Sharding: Complete Hands-On (PgBouncer + Citus). 2 h 36 min and 20 chapters: one PostgreSQL loaded until it breaks, then four shards behind a router you write, PgBouncer and Citus, with the 139-page slide deck. The first 15 minutes are free to watch.
New: live weekend classes, batch starts Saturday 7 Nov
All live classesInterview prep packs
Separate from the course, and sold one by one or all together. Each question comes with a full answer and a diagram.
