I am an independent contractor helping small businesses set up their Data Analytics pipelines among a few other Data Science related stuff that they need help with. Contrary to most of the other attendees here, I’ hold a degree in Economics & I’m very interested in seeing the implications of Deep Learning in my field.
I like to keep self introductions short but if you’re willing you can check out my Medium profile, where I’ve recently started sharing updates and other information from my field. You can also follow me on Twitter to be instantly notified of when I share a post.
It looks like the time of the year when our favourite little icons of the fast.ai family from all around the world become hyper-active on the forums and we get to talk a lot about Deep Learning & beyond
I’ve graduated this year from a CS degree, if you’re curious, spoiler: it didn’t make me a better coder and I was fortunate enough to get a “dream job that I couldn’t have dreamed of” thanks to the awesome community and the course by Jeremy, Rachel and Sylvain
I’m really also hoping to take the course in-person by which I hope to bring the knowledge and experiences from the SF World onto the forums and back to my country too. My old classmates would remember “Making NNs Uncool again” from even then a very awesome course
About what I do/aspire to do:
I try to Kaggle and I’m aiming at becoming a 3-Tier Kaggle Master this year (Moving close to it a little bit every day)
I also try to contribute to making the ML field a little uncool-er for the community by sharing 2 interviews per week with Practitioners, Kagglers and Researchers on a Podcast called: Chai Time Data Science, Audio Link
I’m really excited to take Fastai v4 and continue becoming a better Kaggler/Coder/Blogger (In the shorter: 5-10 year) term and hopefully apply these ideas to challenges that could help the world at scale-agritech (Longer->10 year term)
During the course, I’ll work on 1 blog post per week based on the lecture and 1 based on lecture notes although I suspect it might end up being more than that number.
Really excited about the course and I look forward to meeting all of the fast.ai family members from all over the world on the forums.
My name is Robin and I am a french image processing engineer (a young one, I started 3 months ago). I have been using fastai for over a year now, starting with the MOOCs on youtube, and later with the library when I started my internship. That is also when I started posting a lot on the forums, as many parts of the library felt very unnatural to me. So at first I mainly asked questions, but as I am an impatient man I started answering them myself by going indepth into the source code and testing the different functions to understand their behavior. Therefore I have been trying to answer as many questions as possible for some months, before having to focus a bit more on my work.
Lately I have not been using the library much, as I wanted to have a better grasp on what is happening behind the scenes when training. I considered starting using v2 for a moment, but it was still on early stages and I needed something quite robust for my job. So I transferred to pytorch lightning, which is good for many things and bad for others, but at least I am forced to make things myself. But this is the opportunity for me to get back to fastai2 now that it is more stable and complete, though I’ll probably finish my current project without it.
Speaking of which, I am currently working on a project that aims to grade breast cancers based on histological images of them to help doctors choose the right treatment. Since I started working on DL I wanted to make something that felt useful with it so I’m delighted to be working directly with a hospital to maybe help save some lives.
I am very grateful to be part of this wonderful community, and I’ll certainly be trying to get active on the forums again.
Best,
Robin
EDIT: I have also been a nbdev user since release and I am delighted with it, it is amazing how it helped my workflow ! Thanks also for this !
I’m Vijay. I work in chat bot development and also occasionally on building data pipelines.
I read all the intros before mine and I’m in awe of the background of the people that’re taking the course. It felt like whole Earth for me. Also all the people here, if built a company, that would be incredible. Incredible in the sense of what all problems that company could actually work on. There’s geo-spatial analysis, health care, natural language, economics and what not.
Though I’ve been around the forums since the Part 1 2018 I was never able to complete the course to a full a extent, let it be the timings or the lack of motivation. For the past six moths, I had been following some good people and their work from the extended fast ai community on Twitter. I was astonished at what they’re achieving. Looking forward to stick to the course this year and may be inspire someone towards the completion.
P.S: Getting my new laptop this week, an added motivation for the course
I am Joan and I’m working as a Data Scientist in Hospital del Mar in Barcelona. Although my main work is related to other stuff I am pushing here to develop a DL based clinical related tools (with fastai!) in the day to day care. We are mainly focused in vision and histology slides concretely, but we plan to broad our spectrum when things start to get smoother.
My DL journey started a year or so ago and I had little experience in coding (and python) but I am struggling to improve my programming skills as fast as possible. Fast.ai v1 library help me a lot here since it was super easy to implement SOTA. Now I am playing with fastaiv2 (with @muellerzrcourse and help) and I find the update awesome!
As you can imagine, I am now very interested in the medical imaging module, deployment but also in importing PyTorch code to fastai. However I am really open to new and exciting projects in other DL areas!
Look forward to learning fastaiv2 and hope I could take most out of this course!
Cheers,
Joan
I am Pablo, and my professional trajectory is a bit odd. I have a PhD in biophysics, and then made a postdoc in Copenhagen in neuroevolution (design of neural networks using evolutionary techniques, mainly an algorithm called NEAT) and also in mixed-initiative, where the design process is shared between humans and algorithms. After this I came back to Spain to join the company of a colleague. We have mostly done technological consultancy projects, and I specialize in NLP (I try, at least!).
I found Fastai in early 2018, when I was preparing to start working with Deep Learning. What really sets Fastai apart for me is the focus on top-level results from the first lessons and, much more importantly, the confidence that the Fastai team spends a lot of time testing all the many, many papers coming out in the field, and then incorporating (super quickly) what really works. The downside is that Fastai code itself moves very fast! I started using v0.7, then had to adapt to v1 and now we are moving our code to v2 (we are working on a devilish problem, so every update and little trick helps). So far, a reasonable price to pay
I also followed last year’s course, and I was very excited about part 2, where I gained a more solid understanding on how Fastai is built (really important if you want to change stuff!)
I would really like to think I am contributing to this community, although I am rather humbled by so many of you doing amazing work here and sharing great contributions. So I am specially thrilled to have been invited to follow the 2020 course!
Regards,
Pablo
EDIT: You can find me at twitter at @pablo_gps. Note that I post both in English and Spanish, though.
I plan to get involved in NLP heavily starting this year, working on both English and my native language Vietnamese (which was a goal of mine when I first started doing deep learning but never got the time to work on it). I have read some papers and currently doing Rachel’s NLP course. I plan to implement a few projects using the new version 2 of the library (I have heard a lot of good things about it!)
Overall, receiving Jeremy’s invite for the 2020 fastai course is a huge honor, since the most useful and practical knowledge about deep learning I have ever known comes from Jeremy’s courses. And another good news is that I am selected for USF’s Data Science Master program starting this Summer so I am extremely psyched to move to San Francisco and get to work with this year’s cohort and the fastai community there!
I’m super-impressed by all the bios here of people doing great things with DL. To add some balance there is also room for people like me who are still exploring ML and DL without working in the field. I have been following on-line courses for the last couple of years and am now completing an in-person data science boot camp that I won in a Kaggle competition (thanks to fastai).
What would I like to get out of this course? I’d like to become comfortable working in v2 - I’m already trying out the tabular notebooks - and start doing original projects with a geospatial/environmental focus. Apart from DL, I am also a big fan of data viz.
Looking forward to the new course, although I’ll be on a plane during the first class, flying into Las Vegas. That’s near San Francisco, right?
Hi everyone, I’m Miguel. Currently I’m a PhD student of Geophysical Sciences, working mostly with satellite data and weather models/reanalysis data on applications related to forest fires. I recently published an article on a deep learning approach for mapping and dating burned areas using satellite images (https://authors.elsevier.com/a/1aN0a3I9x1YsQn). I’m a fastai user from the very start, I think I’ve watched all the versions of the course, or at least back to fastai v0.7. It’s always a good experience, there’s always something new to learn and the community here on the forums is amazing!
Hi everyone. My name is Patrick and I am a PhD in Computer Science with core interest in mathematics and computer vision. I have been using fastai since the first day and I am still exited. Shortly, there will also be a paper where I used fastai to learn models with synthetically generated data. As soon as it is published I will report. I shortly started to look into fastai v2 and it looks really promising. A lot of progress has been made since v1. Also I attended all courses till now and are really looking forward to the next course and one more thing: thank you very much for the invitation.
I am a geophysicist based in Australia mostly working with seismic data.
I have a coding background in C, VB6, Java and more recently Python. I started learning ML in 2016 using scikitlearn then tensorflow and through fastai pytorch. Fastai has been a great course and community to learn with.
I spend a lot of my free time either working on DL competitions or trying to code different project ideas.
In my DL journey I started with CV then time series data then NLP and now back on CV problems.
I haven’t used fasatai 2 much yet and am excited to get stuck into it
Hi Everyone. It is an honour to be part of the fast.ai community. A little bit about me. I have a systems, security and infrastructure background. I have no formal qualifications in CS or AI which is what attracted me to fast.ai in the first place. I am self-taught in anything IT related and found myself gravitating to AI in the last 4 years. I see many possibilities to apply this technology to assist the partnership between humans and machines. I regularly use the techniques learnt from fast.ai in my day job as a Systems Engineer at Dell Technologies.
The AI landscape is moving very fast and I get anxious trying to keep up. I hope this part can distill some of the new concepts and architectures that have emerges in the last year or two.
I am Data Scientist from Brno (Czechia). In January, I have returned back to research after two and half year working for industry (Google + an early stage startup).
I am mostly interested in applying ULMFit approach to genomic sequences and (as a hobby) to Czech/Slovak language. I am still collaborating on time series projects (wondering when DL is good idea and when classical statistical models serve the best).
Cannot wait to switch to fastai2 (most of my colleagues are still on TF1). Contact me if you are Czech/Slovak, interested in genomics / time series applications.
Hey Fastai gang!
Getting this invite to Fastai 2020 “Avant première” is a big honor!
My name is Antonin Sumner, I’m a french architect, I’ve practiced 6 years and I have stopped practicing 18 months ago to study full-time mathematics, programming and ultimately our new friends neural networks.
I’m fascinated by the ability of Neural Nets to explore and approximate the latent secret features and functions of the world.
I chase this vision that Neural Nets and architects will ultimately work hand in hand and allow the architect to work in a more organic way. A little bit more as a sculptor and a geneticist, playing with shapes and experimenting with the DNA code of a building.
Also, I’m convinced that this field will help tackle the high dimensions problems of nowadays urbanism and city transformations that we so desperately need.
Currently finishing Part 2 of the Fastai course, I can’t get enough of this, digging deeper and deeper into the mechanics of deep learning, this is just one of the most amazing thing I’ve been offered (well self-offered) to do in my life.
I took part in the recent Kaggle competition ASHRAE - Great Energy predictor, where we had to predict 2 years of energy consumption on a set of anonymous buildings provided 1 year of data.
Even though this competition was more ML techniques oriented, this has been an amazing journey, learning to handle those gargantuan time-series, learning to use extensively pandas and NumPy to visualize, clean, normalize and augment the data all in an efficient way because those datasets were absolutely huge.
I’m currently toying with various types of GANs to find a way to create images of buildings in a creative way, hand in hand with the NN.
Ultimately I’d love to work in a team that explores the cross-road of architecture and data science and the creative process. I know there is more and more frim working in this field and I’m trying to build a suite of prototype projects to be able to integrate one of those teams.
Hello, my name is Darek, I live in Warsaw, Poland. I love spending time with my wife and daughter, I like my work that is about figuring out how to apply technology to improve the way my company operates, I enjoy practicing kung fu, and my latest crush is deep learning I started falling in love with DL 2-3 years ago via Andrew Ng and Coursera, but the serious relationship started only a couple months ago when I discovered fast.ai. I’ve been through part 1 and the NLP course, I work on a personal project that is about applying NLP to understand Polish language (more to come!), and I’ve made a detour in the last couple of weeks to practice on Kaggle (here’s a link to a writeup on my second medal). I’m excited to be here!
I am Andreas a Machine Learning Engineer with a strong focus on NLP @reply.ai.
We use Deep Learning to build smart agents that improve customer support.
I owe fast.ai a lot of my deep learning knowledge and I am happy to be part of this version of the online course. This community is so awesome. One of the things I am especially proud of is the collaboration with @quan.tran. We extended he fast.ai text module based on ULMFiT to accept structured meta data as additional input. Check out the article on Towards Data Science : Next Best Action Prediction with Text and Metadata: Building an Agent Assistant
The future is already here - it is just not very evenly distributed. - William Gibson (my favorite sci-fi author)
I am Ronaldo from Brazil. I do mostly Data Cleaning and Automation scripts at my 9-5 job. I mostly do ML and Deep Learning for fun. Still a Newbie in Kaggle Competitions but it’s where I have the most fun.
I did fast.ai 0.7 and 1.0 Part 1 and use it succesfully at least to Classification Problems. Last time I didn’t quite follow Part 2. This time I hope to rock it!
