Yad Faeq

Yad Faeq

Building All Day Everyday

 

Hello. I’m Yad.

I love computer science and literature.

I do applied research in machine learning on several scales of depth. I use the wisdom I’ve gained on the way towards what’s coming next (it’s not Winter),
the future of AI.

I am looking and open for exciting projects and research. Ping Me? I love what I do and I do it best with people who are passionate about their projects. I also work with some amazing advisors :), if not me, then I can ping friends to advise.

Here is my Statement of Research Interests

 

 

Twitter Blog GitHub Email

 

What Else?

 

My decision to work and focus on machine learning:
This decision comes after realizing that I can make the very best use of different fields of computer science that I’ve enjoyed working in for years.

    During my work journey, I’ve enjoyed programming (<3 graph algorithms), network technology (Switching & Routing), to game design (A.I.), web architecture (Microservices) and engineering distributed software.

Current Status:
Pretty much reading and studying several papers in a week, step by step from the question it’s asking to how far I can get to implement these papers.

    Previously, I use to lead an awesome engineering team, we built internal products to an online school, to an end to end video live-stream service, many other education related resources.

 

Current Projects:

OpenAI Challange 2 (Description2Code)

“Build an agent to win online programming competitions. A program that can write other programs would be, for obvious reasons, very powerful.”

Artificial Intelligence Open Network

AI•ON is an open community dedicated to advancing Artificial Intelligence by connecting researchers and encouraging open scientific collaboration.

Ask Me About This One!

WORK HISTORY

2016

Applied Researcher within Deep Learning / AI

I do applied research with other research teams. My research interests are mentioned on here.

  • Implementing models from papers.
  • Working on Neural Machine Translation models
  • Improving upon Seq2Seq Architectures
  • 2015

    Research Software Engineer

    4th n Town

    I collaborated with a team of data scientists to build an Engineering Assistant using different approaches of machine learning, from linear models to deep learning.

    Here are the models we built:

  • Matching Issues with commits to avoid duplicate work.
  • Predicting labels for issues, whether: Bug, Feature or a Chore
  • Generating commits from code.
  • Collecting Daily Stand Ups and summarizing them.
  • My work came in:

  • Extracting data from Github open dataset.
  • Deploying models into production as Microservices.
  • Building ChatBot in Ruby, to Connect with Slack, Trello, Github.
  • Issue Dup-Detector Demo Issue Labeler Demo

    2016

    Computer Science Instructor

    Codementor

    Teaching computer science and introduction to machine learning on Codementor. The students were either post-docs from other fields getting started in Machine Learning, or currently students in Masters programs getting familiar with the industry.

    Topics taught:

  • Fundamentals of computer science.
  • Introduction to deep learning.
  • Work done:

  • Preparation of data and testing existing models with the students.
  • Deployment of models for working models locally.
  • 2014

    Tech Lead / Fullstack Engineer

    Draper University

    I previously use to work at Draper University, a school for entrepreneurship. I worked with an awesome team to build the following products:

    Draper University Site

    Draper TV

    Draper U Online School

    2013

    Web Architect

    Draper University

    I built a video streaming service end to end to serve 17 Terabyte of videos.

    Using AWS services to it’s limits:

  • S3 with Elastic Transcoder right after video uploaded.
  • Lambda Service with EC2 instances to fetch the videos to application side.
  • CloudFront with SQS to cache and expire videos streamed.
  • 2012

    Software Engineering Lead

    Sunrise Networks

    Sunrise Networks is a technology solution firm founded in Germany, I worked in the branch based in North Iraq. I lead the software development team to build enterprise software for clients such as the Kurdistan Region Gov.

    2011

    Founder & Software Architect

    Easy Route

    Easy Route was a short lived startup that I founded during my school years working with Microsoft with the help of my Professor (Dr.Hemin Latif). It’s purpose was to find critical and faster routes using close friends routes that shared on the network. Mostly to report traffic accidents. Now, it’s a little tiny feature on Nokia Maps.

  • We introduced the use of GSM / GPRS data for geo-location without real internet coverage in developing countires.
  • Built end to end hybrid graph algorithm to offer routing using GSM / GPRS data.
  • Used C#, ASP.net and lots of real graph algorithms.
  • EDUCATION HISTORY

    2013

    Draper University

    Draper University

    My official introduction to Silicon Valley through Tim Draper’s experience. I had the pleasure to learn from amazing founders and venture capitals such as

  • (Shawn Fanning – Napster), (Yukihiro Matsumoto – Ruby Language), (Nathan Blecharczyk – AirBnB), (Robert Metcalfe – Ethernet / 3Com), (Nolan Bushnell – Atari), (Aaron Levie – BOX), (Elon Musk – Tesla / SpaceX), (Josh Buckley – MinoMonsters).
  • (Steve Jurvetson – DFJ), (Dave McClure – 500 Startups), (Michael Seibel – SocialCam / Y Combinator).
  • Studied Futurism, Blockchain Technology, Design from IDEO lab, 3D printing from Adobe, Finance; Wall Street Style, and Survival Training.
  • 2013

    Bachelor of Science in Computer Science

    The American University of Iraq, Sulaimani

    The classes I loved; HTTP Protocol & Web Architecture, Algorithms Ground Up, Advanced Network Security, Research (Graph Algorithms), Capstone in NLP.

  • ACM based curriculum.
  • I didn’t just study Computer Science :), I Started 3 student groups around Computer Science here.
  • Built my first startup out of a research project here as well.
  • 2012

    Game Design

    Columbia College Chicago

    😐 , nope, game design wasn’t the right one, dropped out back to CS degree at AUIS, though I learned about these topics in the meantime over there:

  • AI; Basic of AI algorithms within Video Games.
  • Literature; Studied ‘Hero With A Thousand Faces’ by Joseph Campbell.
  • Interactive Art in Media.
  • 2011

    Comptia+ Certifications

    A+

    Network+

    Security+

    Cisco Certification

    ICND1

    Extra

    2013

    Web Developer

    Beyond the Bombs

    Beyond the Bombs is a transmedia project redefining how the world views, understands, and connects with the Middle East and North Africa.

    BtB Site

    2012

    Founder & CTO

    KnowFund

    Knowfund was a startup that I founded during my senior year of University. It was a micro-crowdfunding platform & mobile payment gateway for Developing countries, such as Iraq. Not on the radar anymore, it’s open-sourced on my Github.

    KnowFund

    2010

    Design Editor

    AUIS Voice

    The AUIS Voice is a publication of the students of The American University of Iraq, Sulaimani. I was a member of the founding team and later a design editor for the Newspaper. Truth to be told, it is the first Student Independent Newspaper ever in the history of the country.

    Machine Learning Frameworks

    Tensorflow

    Keras

    Theano

    NLTK

    spaCy

    Gensim

    Torch

    Backend Web Frameworks

    Flask: Python

    Ruby On Rails: Ruby

    Sinatra + Rack: Ruby

    ExpressJS: NodeJS

    Django: Python

    Frontend Frameworks

    ReactJS

    BackboneJS

    Databases

    Postgresql

    MongoDB: NoSql

    DynamoDB: NoSql

    Redis: NoSql

    TitanDB: GraphDb

    Honors & Awards

    2016

    Top Writer 2016

    Quora

    “The best answer to any question”. That’s what the goal is on Quora and I tend to share my part.

    2014

    Microsoft YouthSpark Advocate

    Microsoft

    YouthSpark Advocates are young leaders who have been selected to represent YouthSpark through online and offline activities. Advocates become subject matter experts in Microsoft YouthSpark world.

    2013

    AMENDS Delegate

    American Middle Eastern network for dialogue at Stanford

    AMENDS is a student initiative at Stanford University that brings together the most promising youth change agents from across the Middle East, North Africa, and United States to learn from each other, connect with global leaders and resources, and share, through TED style talks, their ideas and experiences with the world.

    2011

    Founder of ACM Chapter in Iraq & Ambassador

    Association for Computing Machinery

    ACM, the world’s largest educational and scientific computing society, delivers resources that advance computing as a science and a profession.

    2011

    MSP & Iraq’s 1st MSFT Student Ambassador

    Microsoft
    2012

    1st Place on locals winner Imagine Cup

    Microsoft
    2011

    2nd Place Locals winner Imagine Cup

    Microsoft

    I ramble about about technology here n there

    2016

    Foresight Institute Artificial Intelligence for Scientific Progress

    Palo Alto, CA

    With the awesome team AI 2016 Foresight conference, We presented a proposal for building an agent with Deep Reinforcement Learning to learn and generalize over different environments. This agent to function as a junior researcher, to be used in labs for performing sequential experiments.

    2015

    Software Engineering Daily

    Data science is a broad topic with numerous sub-fields such as data engineering and machine learning.

    Yad Faeq returns to the podcast to discuss data science at a high level, and rescue Software Engineering Daily from the threat of the hype vortex.

    Yad is a software engineer, currently working on machine learning applications. Yad first appeared on Software Engineering Daily on Episode 0, the inaugural show.
    Questions:

    • What algorithms does a data scientist need to know?
    • How would you architect a machine learning system for weather analytics?
    • What is deep learning?
    • How has Kaggle changed data science?
    • As machine learning comes further into our lives, does our world become more predetermined?
    2015

    Software Engineering Daily

    In Episode 0 of SE Daily, Yad and Jeff give a prologue to many of the topics that will be covered in JavaScript Week.

    • Are we beyond the era of monolithic apps and frameworks?
    • What is the switching cost if you want to move from one JavaScript templating language to another?
    • Why does Uber use both MongoDB and MySQL?
    • Why does Quora only use MySQL?
    • What does the Node.js event loop solve?
    • What is the difference between Microservices and SOA?
    • Why is Docker relevant when we have had virtualization and containers for a long time?

    below are some old old old stuff : ), but still worth sharing

    2013

    AMENDS 2013

    @Stanford

    Talking about project “Knowfund“, my project at the time. (created a mobile payment gateway & a micro-crowdfunding Web app as a part of the solution)

    2012

    Imagine Cup 2012 Finals

    @Erbil Hall

    Pitching “Easy Route“, my first startup with my team on stage in front of judges and an audience close to 1k folks! (Terrifying, yet awesome experience). Also this is recorded by my awesome professors (love & hugs to them).

     

    Draft Of My Research Interests (Work In Progress)

    1

    Sequence to Sequence + New Architectures

    Improving Upon Sequence to Sequence Learning

    RNNs and Sequence to Sequence Learning is a bit of a broad topic, to be more specific:

    • Learning the representation of programs and generate for multi-tasks. This is a problem that can be tackled with seq2seq
    • Applying machine translation techniques to tasks other than human language translation. There have been amazing results in the past few months following up with (Bahdanau, Cho & Bengio 2014 from 2014. Different use-cases of attention (Luong, Pham Christopher & Manning 2015) and in Sequence to Sequence learning is an exciting way to improve upon these models for improving upon RNN architectures.
    • Also, working around architectures to improve these models is exciting, an example of which improves the LSTM encoder decoder Sutskever, Vinyals & Le is this approach from (Jaitly, Sussillo, Le, Vinyals, Sutskever, Bengio 2015)
    • There are loads of roads to take at this point, my goal is to narrow down the list of architectures to the several domains I will be applying to.
    • Exploring new architectures with attention towards problems that regular seq2seq wasn’t equipped to tackle. Recent work by Ilya Et al(Seq4) is a good example of the type of architectures that I’m exploring.

    More details coming soon.

    2

    Deep RL + Learning To Learn

    Deep Reinforcement Learning in Hard Sciences (natural sciences)

    Hard Sciences is abit of a broad topic, to be more specific:

    More details coming sooon.

    3

    SLAM + Logistics

    • Last 10 mile problem for autonomous vehicles.
    • Mapping narrow areas, autonomous parking and navigating inside cities
    • Engineering problems! Models need to be fast and responsive enough.
    • Vision, stitching scenes and exploring alternative sensors.

    More details coming soon.

    4

    GANs

    Generative models

    • Using video games as an environment for generating data / learning
    • GANs in education, as meta-agents for instructors.
    • Building stable outcomes on the current status of generative models.

    More details coming soon.

    5

    Ethics + Democratization

    Ethics Of AI:

    • A lot of folks introduce Ethics as a post AI event, I **strongly** believe that we first need to fix the ethics of the community before, otherwise we won’t be able to reach the kind of AI that is sentient (The Master Algorithm)
    • Building tools that work as a catalyst and ease of access. Such as [Keras]()
    • Lowering the barrier of entry of other folks to learn about machine learning. Such as the work of [FastAI](), Stanford Classes [CS231n]() [CS224d]()
    • things like OpenAI, AI-ON.Org
    • We need more [open-ness](https://twitter.com/AndrewYNg/status/784182224954335233)
    • Now to the technical side of EOAI:

    • Building interpretable neural networks architectures, that are justifiable [source](http://nautil.us/issue/40/learning/is-artificial-intelligence-permanently-inscrutable)

    More details coming soon.

    6

    Other experimental work:

    • Restricting NP hard problems in sandboxes and solve for Optimality with machine learning. i.e: Optimal TSP solutions.

    More details coming soon.