Just updated my website for the first time in a longgggg time (4 years to be exact!)

Its great how much SEO I've built up on the old one, but I think its long overdue for a fresh coat of paint (and description), so here it goes...

My motto while building this was "make it simple, but surprisingly complex... WIDGETSSS"

I updated basically everything but here are the features I'm most proud about:

- A spotify widget to the bottom left (exposing all my songs)

- A location widget that receives a city & weather update every 4am from my phone

- A notebook widget to check up on my posts (had to add an rss feed for this)

- Lastly a github widget to see my last commit

- I had to finagle a good way to get static & dynamic widgets working because fetching my large RSS feed was taking too long (I eventually settled on fixing it on my notebook side, but I hope Next13 will be compatible with emotion ui soon...)

I'd like to add a couple more widgets in the future (maybe something that interacts with something in my room? DM or email me if you have any ideas) but I'll get to that after college apps

Getting to make something that wasn't super complex after being (very slightly) "burnt-out" due to college apps was super fun.

Updating my website for the first time in 4 years!

Have you ever been unfocused at home, or feel depressed? With the pandemic, millions have faced many hardships with their life spiraling out of control. To combat these problems, we developed TARDIS, a neurotech headset that's able to let you take control of your life by monitoring your state of mind using AI and help you stay focused. We allow you to detect what state of mind you're in and gather statistics about your brain, and choose to nudge your mind into another state, like being focused or help with depression. You can learn more about it at the website, and check out the white paper or pitch deck on there!

This year a couple of friends and I founded TARDIS, and applied to the Conrad Challenge, an international innovation competition. Over the past months, we've won the competition, getting first place internationally as well as a diversity award, and made it to the Conrad Summit, where we were able to speak with investors, listen to talks from our peers, and hear feedback from judges! It was an amazing experience, where we got to speak to so many people, and I learned so much more about business than I ever thought I would know!

The summit was a great experience, where I was able to connect with high schoolers and mentors from all around the world and receive feedback on our product and marketing. I can't wait to take this forward in the next few months, trying to secure funding and designing our MVP!

A cool new side project, and my experience submitting it to Conrad Challenge and winning!

Yay! The Contra Costa Science Fair just ended, and I ended up getting 3rd place! I had a ton of fun on my project, and I learned a lot on neural networks, especially generative semi-supervised architectures. Check out the video, or the explanation below to learn more about it!

A little bit about my project is that, Leishmaniasis is a neglected tropical parasitic disease that kills millions of people in poor underdeveloped countries. Those that suffer from this disease cannot afford expensive medicines as the R&D cost is too high leading to limited commercial development. The goal of this project is to utilize AI to build an open-source neural network to generate novel molecules for leishmania. Utilizing AI for drug discovery would decrease R&D costs and being open source will allow development or other neglected diseases.

The target chosen was Methionyl-tRNA synthetase, which plays a key role in protein synthesis in the parasite. This target also has a lot of prior structure activity data, which allows for training the model.

The network used was adapted from GENTRL (Generative Tensorial Reinforcement Learning), outlined in Dr. Zhavoronkov's paper, where they used GENTRL to successfully create a novel molecule for DDR1 Kinase. The model uses a variational auto-encoder to determine features in the data, and then uses reinforcement learning to tune the decoder based on pre-trained self-organizing maps. The rewards I used were log P, synthetic accessibility, and log IC50 values. The datasets used were MOSES leads, general ligase, and specific target data.

After training, the model was able to generate promising novel molecules. The final structures had a maximum potency of 93%, and a 2.8 accessibility score, implying significant synthesizability for a bioactive molecule. I showed the final output to an expert discovery chemist, who said the molecules looked feasible to create, and were sufficiently complex to warrant bioactivity.

Woot woot! You can check out my code at the GitHub repository, or download a pdf of my board! I hope to either build on this with testing, or make something even cooler next year!!!

Using neural networks to develop novel molecules to target Leishmaniasis. A documentation of my science fair project!

LeishNN: AI based Drug Discovery for Leishmania

Over the past few months the TeachAids team and I been developing a way for children to learn about covid19 easily and have a lot of fun while doing it! The end result was CoviDB 4 Kids. While interning at TeachAids, a team was put together to develop a way to solve the problem that new information was mainly being targeted towards adults, while children were kept in the dark. After brainstorming a few ways to go about it, we settled on a website where kids of all ages could easily watch curated videos on the topic.

After months of development and input from our fabulous designer, the end product was done! Built with Next.js, Theme UI, and Rebass, CoviDB 4 Kids uses a kid friendly UI to teach kids from 3-12 about covid19! It was really fun, and an amazing project to work on, and I got to learn so much about how to think about user experience, as well as developing a website professionally. 

I just wanted to take a second to say a big thank you to everyone I worked with there, as its always been fun to work with them and talk with them on our calls!

My story on how I helped TeachAids in developing CoviDB 4 Kids,  a COVID-19 (a new coronavirus) educational resource database for kids!

&nbsp;A few weeks ago, I got accepted into an internship program for [TeachAids](https://teachaids.org/)! So far, its been an amazing experience where I get to work with many people and develop my skills! I've leared so much since I've started, from editing videos (with premiere!), to learning how to use the twitter and google drive API's ( the latter sucks :\\ ). Its also one of the things that I've been wanting to do for a long time, because I usually develop projects solo, and I want to get the feel of working as a team to create a finished product!

&nbsp;So far, I'm planning on doing a full time summer internship, and to check out the things I'm working on, look at [CoviDB](https://covidb.org), a great way to research on covid-19, the [speaker series](https://covidb.org/Video), and stay tuned for Story Wall, a great way for people to share their concussion stories!

A fun internship where I'm working on many cool things, and learning about new technologies!

After working with Go for a while, I decided to hunker down and write a production level CLI for fun. Lo and behold came templater. Templater is a really cool CLI for storing, loading, and creating code templates. Its as easy as 

and then you have created a template! Then with simple GitHub OAuth, you can upload the template easily, and view it on the website at https://tmpltr.now.sh or just download it through the CLI. The best part is that when uploading, it saves the file structure and README.md, so that you can check out some basics of what it is before you download it. Another thing I tried to do with this project is proper git etiquette (PR's, Versions, Badges, Unit Tests etc.) and it turned out pretty great, where I learned a bit about shields, and how to configure GitHub actions. Check out the GitHub at https://github.com/neelr/templater and be sure to leave a star! Also, please try out the CLI easily with

1. `go get github.com/neelr/templater/cmd/plate`
2. `go install github.com/neelr/templater/cmd/plate`

A fast, easy way to organize and share code templates.

*Example Anime Music AI Clip:*

<audio
        controls
        src="https://drive.google.com/uc?export=view&id=1OGHKyOJIO00pH471kPcXbm93hFIJ7erR">
            Your browser does not support the
            `audio` element.
    </audio>

So this week I decided to get down and finally create a cool project with AI! After thinking about something that would be cool to make, I decided to try and make something that I really like, anime music. This wasn't completely new territory for me as I have experience writing neural nets, and working with time series data, and had an idea of how to approach it.

For more in depth comments and understanding, I recommend reading my Jupyter Notebook on GitHub

Before I started to build the neural network, I decided to look around to see if there were already any good anime music datasets on Kaggle, and lo and behold, there were none. So I decided to think about what data would be suitable for a project like this. I laid out my criteria:

Has to be digital, not analog. This is because neural networks won't always be 100% accurate, and it would be very hard to output a sound wave that sounds decent, vs digital data would be easier to work with and way more forgiving.

I would need good anime music for the model to learn off of.

The data type would need to be general and known wide enough so that not as well known anime songs would have to be transcribed in it

After looking at all of that, I decided that the best media format would be MIDI files in a piano format, because it is digital, easy to use, and many people have already converted anime songs to MIDI files. Now the main problem here was scraping data. Though it's not in the repository, I built a web scraper  (keep in mind it wasn't a spider because I wanted to curate every song), and managed to get about 130 songs! These can be found on GitHub or on Kaggle. Next came parsing and training the model.

Lucky for me, there was already a MIDI file parser in python (music21 curated by MIT), so I was easily able to parse the files and get the notes. After that, I decided to take the routes one at a time.

For v1 I decided to only scrape the files for notes and chords, that way it would be easy to test if it had a possibility of sounding good.

Next for v2 I decided to scrape the rests as well as the notes and chords, increase the amount of data (previously 88), and push back the look-back (will get to that later) to 200.

Lastly for v3, I decided to go all in with including tempo, keys, and key signatures. This ended up increasing the amount of notes significantly which affected the neural network architecture.

After taking all of those routes, I started to train my neural network. The way I envisioned a person to do this, was take a look at the past notes, and infer what the song should be. To do this, I decided to use LSTM layers (Long-Short-Term-Memory) which basically store and cycle through past values. I also inputted 100 of the past notes (upped to 200 in v2 and v3), to get 1 note out, which is what I suspected would be the easiest for a neural network to do (because it would be feasible for a human to do). For my actual neural network architecture, I'll put an image below.

I bet now you're just like, SHOW ME THE MONEY!!! So here are some samples (more can me found on the GitHub or on the Google Drive)

**Version 1**
    
<audio
        controls
        src="https://drive.google.com/uc?export=view&id=12Ylz8gR38IzQFlm4louzvW1mPKTAbqUr">
            Your browser does not support the
            `audio` element.
    </audio>

**Version 2**
    
<audio
        controls
        src="https://drive.google.com/uc?export=view&id=1O_jbtf1NH0Z77UjTW-pP-ZguyWIAAqwp">
            Your browser does not support the
            `audio` element.
    </audio>

**Version 3**
    
<audio
        controls
        src="https://drive.google.com/uc?export=view&id=1R7yEWolMKyINZXM83eB9LWD4G_RtvzJ6">
            Your browser does not support the
            `audio` element.
    </audio>

In short, this was a really fun experiment where I learned a ton about LSTM networks, and music encoding. Technically, you could run and train this on other MIDI file collections (gaming, classical, hip-hop, etc) music, so I might try to make Undertale AI generated music in the future ( ͡° ͜ʖ ͡°). You can check out the GitHub at https://github.com/neelr/AnimeLSTM, to take a look at the source code! What I want to do in the future (and will link to if done) is make a 24/7 music stream of AI generated music on YouTube so I can easily just play and enjoy some music whenever I'm studying. Leave a star if you enjoyed the music, and thanks so much for reading to the end, and feel free to contact me at neel.redkar@outlook.com!

I visualized and created a 1 hour YouTube video to listen too!

A project where I generated my own anime music using AI!

## What is this?
One day, I decided to learn a bit more about neural networks, and set myself to try and build and train a neural network using reinforcement learning. This is a great start for me because I don't have to start to learn back propagation, and can rely on mutation. So first I decided on a game, and realized that flappy bird was one of the best games to try reinforcement learning with. This is because the rewards are easy to figure out, and you can spawn many birds in one environment. Kudos to Coding train for the [initial base flappy bird game](https://github.com/CodingTrain/Flappy-Bird-Clone).

## Neural Networks
The cool thing that I learned about neural networks during this time is that its simple linear algebra! If you consider the neurons to be a vector, all you need to do is take the dot product of the nodes and the weights, add the bias, then you get the new neurons value! I was so surprised that it was so simple, so I started to create a class, along with a mutation method so that I can mutate each weight slightly so it can learn like in Darwinian Evolution. Bellow is the neural net used for the birds. It takes in the velocity, y value of bird, y value of top and bottom of pipe, and x value of pipe! The output is 2 numbers, and if the first node is larger, it jumps. That way I don't have to implement a soft max activation.  
<img style="width:80vw;max-width:400px;" src="https://images.prismic.io/notebook/ba212d43-c640-457c-a3f0-7cec208417d6_Screen+Shot+2020-05-11+at+3.29.01+AM.png?auto=compress,format"/>

## Training
After building the neural network, all I had to do was pop it in to the birds, so some debugging with the inputs, and it worked flawlessly! The birds were learning fast, and after a few generations, started to play the game without any mistakes! This was amazing because I didn't think I would be able to create my very own AI from scratch, and it's amazing to see something actually make decisions with math.

If you want to check out the project, take a look at https://flappyai.hacker22.repl.co or look at the GitHub at [@neelr/FlappyAI](https://github.com/neelr/flappyai).

A cool game where an AI teaches itself to play flappy bird.... built from scratch with JS Core libraries!

Another cool project that I made which you can check out at https://db.neelr.dev! Its a great replacement for jsonstore, and is my first project in Go. It has easy JSON navigation through URL, and is super fast, so check it out for any small projects where you just need a quick database!

A fun project I wrote that's a simple and lightning fast database written in Golang!

Last year around September, I decided to stoop down, and code a notebook. Basically what a notebook is, is a way for people to write down their thoughts, or anything cool, and make it public so they are help accountable for writing in it so often. The main barrier for me has been that it has been a pain to write in. From the start, I used individual pages, where I needed to write JSX pages, commit them, and watch them build, then I realized markdown was easier, so I started to write in MDX. After a few months of that, I decided that the actual committing part was a pain, and I started to look for a CMS, and decided on Airtable. Airtable is a great database for small projects, but the small writing space and speed became a problem.

With all of that out of the way, I attempt to solve all of my problems with this new improved..... Notebook v2.0! Basically what I changed here, is instead of sloppy Next.js code (trust me, you can take a look at the commit history), I improved it with Next.js, Theme-UI, and Rebass, for a nice clean UI and a minimalist theme. Then from there, I ventured (not far) for a new CMS. What I was looking for was anything where I could :

And that's when I found Prismic! It's amazin, because its free for one person (haha where being alone is helpful), has a great user interface, and has a content system which links to an easy to use database. This was perfect for what I was looking for, and decided to not only make a switch in CMS, but change the design a ton. With the ability to add tags now, I decided to add a tagging system, where I can have tags related to releases and for other types of things (design heavily influenced by Lachlan's and Mingjie's Notebook styles), so its easily searchable so you can find similar content. Lastly, I wanted a way to see how many people actually read it (most likely 0), but didn't want to track IPs or have a view count, so I decided on a star system! Basically, anyone can star a post, and it'll increment the counter. It's hosted on goJSON one of my new Go projects which I'll probably write a post about sometime. You can try starring it at the top or bottom, and it helps me know you liked it, or if anyone reads them (it mostly doesn't matter because this is to improve my writing skills)!

#project

MyWebsite 2.0

TARDIS - The Conrad Challenge

LeishNN: AI based Drug Discovery for Leishmania

CoviDB 4 Kids

Starting at CoviDB & TeachAids

Templater

AnimeLSTM

AI from Scratch: Flappy Bird

goJSON

Notebook v2.0

.css-4cffwv{box-sizing:border-box;margin:0;min-width:0;display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;}#project

MyWebsite 2.0

TARDIS - The Conrad Challenge

LeishNN: AI based Drug Discovery for Leishmania

CoviDB 4 Kids

Starting at CoviDB & TeachAids

Templater

AnimeLSTM

AI from Scratch: Flappy Bird

goJSON

Notebook v2.0

#project