The Rise and Rise of FastAPI

The story of FastAPI is, oh,  it's like, so many parts. When I started working on, like, actually building  FastAPI I was just trying to solve the problem   that I needed to solve in a way that was useful  and efficient for me. I always expected that no   one will pay attention but it actually like, took  off and people started liking it. FastAPI FastAPI   FastAPI FastAPI. It kept growing in GitHub  stars and it was just like this steep line. My name is Sebastian Ramirez  and I created FastAPI. I like to play around and say that I'm a  kindergarten dropout, but it's actually not true.   I'm a elementary first dropout. I was in a, you  know, like regular Catholic school in Colombia,   but I will ask too many questions. I was also  like very curious about many different things,   but I was not really enjoying the way  that the school was working. So I ended   up just like being homeschooled all my  life. Having this mindset of liking to   study and to understand how things work.  I think that probably helped with FastAPI   in that I was just trying to see like is  there any better way to do these things. When I started programming, I was always a  consumer of open source. And then at some point,   it was probably trying to solve a problem that I  had with some open source project somewhere and   figuring out that it was actually for probably the  first time not my, uh, error, but it was actually   something wrong in the tool. And I figured that  I could help it with it and I could solve it.   And I was trying to use the tools that I had at  hand and trying to learn all the tools possible   and then learning the things that I liked about  each one of those. I was actually convinced that   there will be a tool that will have like all the  things that I wanted. I was just trying to learn   what was the right way to do it and what was the  right tool. When I found one that was the closest   to what I wanted, it was called API start. And  this was built by the same author that had built   Django REST framework. the alitor needed to stop  working on that because the alter was working on   starlet that was this new framework that was in  the newest standards and that's had like a much   better performance. I was also doing a bunch of  front end and learning how the front end works   and learning that all these type annotations were  what provides auto completion and inline errors   and I wanted to have that in Python. Then  I figured that there was an alternative to   Marshmallow called Pyantic that was at that time  focused on validation of settings. So I wanted   to combine Pyantic with the things on starlet  and put build something together with that. At   some point I actually went to the to the github  issues in Starlet and I said hey I would like to   build this thing that is pretty much you know  like it's similar to API star that was like by   the same author and I asked would you like to have  this in Starlet and they said like no just like go   build your own thing. So I was like okay I'm going  to build my own thing and then I went and built   FastAPI mixing Pyantic and started and putting all  these things together. It was just trying to solve   the problems that will come natural for building  APIs web APIs using the new H features of the   language type annotations async and weight and a  bunch of different ideas from different tools like   dependency injection. That's how it came to be.  The name fast API, the word fast has actually like   I guess multiple meanings. And I guess one very  important one that people don't necessarily pay   attention is that it's fast to build stuff with  fast API. So you can like get something out the   door very very quickly. It's also h fast in terms  of performance and uh in terms of throughput.   I wanted to build something that would appeal  to myself when I was learning how to use these   things. I wanted to build something that will be  very simple to understand and to figure out how to   use it. I was trying to solve the thing for me as  a developer using FastAPI, not as the person that   builds it, but as the person that uses it. So, it  was a lot of just like designing that developer   experience before actually going and implementing  the things underneath. FastAPI is built on top of   type hints or type annotations. You can use type  annotations to be able to just declare them once   and then get multiple features out of that.  You not only get auto-completion and errors,   you also get data validation, data serialization  and documentation automatically based on just   those same regular type annotations. The fact that  FastAPIs by default is internally async is super   important in like, in in particular nowadays that  when we have applications and APIs that need to   work with lots and lots of concurrent clients or  concurrent uh users when you build with FastAPI   you get an automatic documentation site for the  API. I guess it's one of those things that for   people that when people start working with FastAPI  and they just like build, you know, like a tiny   piece of code and they just go to /docs and they  see this is all happening automatically. That's   normally the moment like, oh wait, I like this.  What's happening here? By now, you've probably   heard that FastAPI is great. I'm going to be  showing you the Python web framework known as   FastAPI. It's really fast. That's why it's called  FastAPI. The first sign that FastAPI was catching   on was I guess GitHub stars. Then FastAPI was on  like I I don't know if Reddit or Hacker News or   something and then it suddenly grew in like a few  days to 500 stars. Then it just kept growing like   a very high speed. Then this was definitely not  expected. FastAPI is currently used by so many   companies like Microsoft, Uber, Netflix, Amazon,  Meta, OpenAI, I don't know, the list just goes   on and on. It's like all the big ones, but also  all the small ones is used by so many people in   so many places. It's also used for many things  that are like super strange and super cool. So   for example the Web Space Telescope, like this  giant telescope in space, all the images and all   the science data that is taken from the telescope  is sent to earth and then distributed across the   world using FastAPI. They are using FastAPI like  discovering new drugs, tracking COVID, tracking   wildfires, handling disaster recovery for the Red  Cross, you know, like so many different things. I'm so happy to see that it's useful to so  many people. I I like, get so excited when   people come and tell me like, hey, we are using  FastAPI for this for like we're using. It was,   you know, like it helped me find a job. It  helped me solve like all these different   types of problems. Every time that I hear someone  telling me something like that, I I get so happy   and so excited. With FastAPI growing, the ideas  that I brought into building APIs with FastAPI,   I want to bring those same ideas and those  same improvements into deploying FastAPI   applications to the cloud. I would like  to empower many people to build more stuff   uh to build more applications. Many  people, many companies, many organizations,   many different types of products that could  be built. I like to help people help others   with the tools. So that is that is in general  what I am trying to do just to help people   be more efficient and uh generate value and  solve problems for all their people around.