I’ll be more than happy to join in!
Hi everyone, my name is Tom Szumowski, from New Jersey, USA. I’m a Data Scientist at Urban Outfitters, previously a research engineer at Lockheed Martin. I have a MS in Electrical Engineering, but I’ve been working with Machine Learning since 2012, and deep learning since 2014. I’m excited to learn more about the fast.ai framework as well as PyTorch underneath since I spend most of my time in Keras.
1 Like
Wayde here … and glad to be back for my 2nd official run through the course.
I enjoy contributing on the forums and to the fastai framework, with a particular interest in anything and everything NLP. As a mentor for a High School FIRST Robotics team, I’ve also been able to pass all this goodness along to students interested in application development and predictive analytics.
If anyone is in the San Diego area, give me a holler.
2 Likes
hmm, now that sounds very interesting. 
3 Likes
Hi everyone! Good to be here with you.
I’m a British software engineer and entrepreneur living in Santa Cruz, CA, and I work with a small team of physicians and researchers to create software to help clinicians make better decisions using structured data.
With almost no data science experience, I attended v1 in-person, and that gave me the tools I needed to get started. While I ended up using gradient boosting to solve my problem, I remain optimistic about the potential of deep learning.
Thank you, Jeremy and Rachel! I’m excited and grateful for the opportunity.
5 Likes
Hello - I’m Ralph from Austin. I’ve been lurking around here for the past year, happy to finally to take a class in real time. I currently work as a data scientist in payment fraud. Thanks to Rachel and Jeremy for making this place so cool and the subject matter so uncool!
1 Like
It is nice to see you here, James !
1 Like
Hi everyone, this is Ignacio from Valencia, Spain. I have a Pharmacy degree, as well as an MBA. In the last couple of years I’ve taken a new career path and have been working in a financial trading startup that uses ML.
I’m interested in getting a better understanding on how to apply DL to financial, tabular/ time series data.
Can’t wait to get started!
1 Like
Hi all! I’m Jeff and I work as an Oracle DBA in the Seattle and Ive worked with data and as a developer for a long time. I’ve never been in a position to use ML in my work but I’m hoping that by focusing on the fastai class, I can get to the point of making a real go of it related to performance prediction and diagnostics.
1 Like
Hi All,
I’m a data scientist working full time at a financial institution (yikes, i know!) My background is in pure mathematics.
Currently my focus is on studying ways to architecture systems of neural networks with independent failure rates. Briefly: it’s clear if you have 2 models that are each 90% accurate and their error-sets are disjoint, you can achieve 99% accuracy on a subset of your domain if you reject any input for which the models disagree. Perhaps disjoint error set is too much to ask, but independence ends up being more than enough to be able to say interesting things. This ends up being very useful on tasks where it’s perfectly okay to be unsure (and call a human expert) but very costly to make a mistake (e.g medical imaging). The trouble is it has proven quite tricky to train two (or more) neural networks to have this very desirable property while still having low loss.
I have several approaches that to my knowledge are novel (but very much in the spirit of many of the ideas developed by Jeremy in earlier renditions of this course!) that I am developing for both NLP and CV domain tasks and have gotten some exciting results already. If you are unsure of what sort of project you would like to pursue in this course I am happy to collaborate, as I am in the stage where after developing a handful of techniques the time has come to test test test across different datasets and problem domains and carefully document the results. Feel free to drop me a line!
Cheers,
Max
2 Likes
Great to see you here too buddy & congrats on all your success!
2 Likes
Interesting stuff !
I would love to read more about it, if you ever write a blog post please post it here in the forums 
Small world! Yes I have read it and loved it. In fact that paper gave me the idea for this project. As per my understanding it was heavily based on embeddings and I thought with ULMFit with some new research add-ons we might get better results.
2 Likes
Hi all!
Excited to be back and pick up on the new v1 API. I took part 1 and 2 earlier this year remotely and landed a job in machine learning as a result. We’ve been working on wildlife conservation with computer vision amongst other things and now looking to accelerate our skills with the latest course.
Jonathan
3 Likes
Hi @christianfjung, you can download the data from here in XML format.
clinicaltrials.gov/ct2/resources/download
1 Like
I suspect you may have found a potential collaborator already!
2 Likes
I’m a long time fast.ai forum reader and fan, and a first time participant! I work on applying machine learning in industry and publishing reference examples based on the work. I have found fast.ai resources deliver exceptional results and are refreshingly practical. I’m enthusiastic about learning to apply the latest NLP techniques. In August I had fun using ULMfit to generate some reasonable poetry based on fine-tuned Billy Collins and William Blake collections. I’m also excited to learn how to use pytorch. Thanks so much for putting the course together!
1 Like
Hi everyone, I’m Sylvain and I’m very excited for this new course to begin.
Thanks to a sponsorship from AWS, I’m a researcher in residence at fast.ai, and I spent the past months working on the new library with Jeremy. I can’t wait to see what each and everyone of you is going to do with it! Before that, I was a math teacher, a textbook writer and a stay-at-home dad. I learned pretty much all I know about deep learning through part one and two of this course, so huge thanks to Rachel and Jeremy.
I pretty much like everything in deep learning, if I had to name a specific point of interest, it would be how to make trainings of models faster and more efficient. In the future, I’d like to finally get time to seriously participate in a Kaggle competition, but there’s probably always going to be more to do on the fastai library first. Not that I complain, working full time on this project while getting paid is kind of a dream come true.
Otherwise, I’ve broken the library more times than I can remember, so if something isn’t working properly and you’re sure it’s not on your end, it’s probably my fault. I try to repair things fast when that’s the case.
37 Likes
Hello everyone! My name is Curtis, and I hope to apply learn how to apply deep learning and ML toward discretionary and systematic trading. I am also interested in how it can be applied toward artistic, productivity, and UVP (unique value proposition) applications. I have a strong background in software engineering.
I have had more success focusing on coding vs. studying. My blog is @ http://beyondbacktesting.com
1 Like
I am a Software Enginner by profession and mostly worked on JavaScript related projects. (Backend and frontend both).
I am from Sri Lanka.
I tried many times to start learning ML and DL. But I couldn’t complete due to lack of interest. (Not blaming those courses)
But I found that fast ai courses are very interesting and not boring at all. That’s why I wanted to follow this course.
My main goal out of this is to build a firewall for facebook which could help users to get rid of online harassment. So, they can express themselves without fear of getting a landslide of hate comments which has no point.
3 Likes