Library / YouTube / sudoCODE

sudoCODE

Beginner · 138K subscribers

System design concepts explained in Hindi and English for interview preparation.

Learn system design with sudoCODE, step by step

77 videos in 5 steps, about 16 h 7 min of watching. Start at step 1 and move down.

  1. 1Start here: the basicsNo background needed. What the main ideas are and why they exist.
  2. 2The building blocksCaches, queues, databases, load balancers: the parts every design is made of.
  3. 3Design whole systems (HLD)Put the blocks together for a real product, step by step.
  4. 4Low-level design (LLD)Classes, patterns and clean code inside one service.
  5. 5Go deeperInternals, trade-offs and how large systems really behave.

Order and levels are picked by us from titles and lengths. The one-line notes are ours. Every video plays from YouTube and belongs to sudoCODE.

sudoCODE’s own courses and series

Playlists the creator put together, in the order they made them.

  1. System Design Primer CourseA series from the creator, in the order they made it.
  2. Low Level Design Primer CourseA series from the creator, in the order they made it.
  3. Design PatternsA series from the creator, in the order they made it.
  4. Microservices - Must know conceptsA series from the creator, in the order they made it.
  5. REST APIs MADE EASYA series from the creator, in the order they made it.
  6. SOLID Principles Made EasyA series from the creator, in the order they made it.

Latest uploads

What viewers say (5)

Public comments on this channel's recent videos. They are not reviews of this site.

  • Mix of both (SD + AL/ML) also very very congratulations for 5year milestone.

    @Manney45· 18 likes

    on

  • This is a very good explanation for RAG systems. But the ChatGPT work in combinations of RAG systems and Generator Systems. GPTs are decoder only type of Transformer models which are pre trained on large amounts of data which allows it to predict the next words based on the Semantic Similarity and Distances between the vectors. This allows for them to generate text/images. In this case, because the query was just asking the LLM to give a response instead of generating some text, you using a BERT was sufficient, in the other case you will need to use the GPT to generate the text as a response.

    @krishnareddy8767· 18 likes

    on

  • The discussion really drifted from the actual system design focus. If the LLM is a black box, there's no need to talk about how it generates responses. The real challenge is how to build the system around it — managing context windows, chat threads, limiting to 50 chats per user, syncing across devices, storing chat history efficiently, and keeping latency low. That’s what a “design ChatGPT” question is actually testing.

    @devmaqsad· 11 likes

    on

  • This is a really good attempt, but it goes into rag like system in a weird half baked way. In my opinion, it should start at the core problem of treating the transformer based text to text generator model as a black box and start with discussions about its input and output. And the challenges regarding scaling on both sides. RAG should come much later, as probably the last functional layer, to specifically solve the problem of bringing in certain world knowledge that the black box model is unaware of. But the systems design should start by first solving the design for the world knowledge that the model already has.

    @TheCuriousPrimate· 7 likes

    on

  • Anything but now wanted longer format content from you,20+ minutes videos

    @abhishekanand472· 2 likes

    on

All videos, channel names, pictures and comments belong to sudoCODE and the people who wrote them. Videos play in YouTube’s own player and link back to the channel; nothing is copied. Numbers and comments come from YouTube and are refreshed regularly. By using them you also agree to the YouTube Terms of Service; Google’s Privacy Policy applies. To have this page changed or removed, tell us here.