Jenndy - Machine Learning Pathway

Certificates June 2020 Certificates Machine Learning Pipeline (NLP) Self-Assessments
1

Concise overview of things learned:

Technical Areas:

-Web scraping a Discourse forum with Beautiful Soup and Selenium
-Learned and built BERT model with Hugging Face (PyTorch and TensorFlow)
-Learned to use classification with scikit-learn on last layer of BERT model output
-Having a technical vision that is concrete yet flexible
-Reviewing code

Tools:

-Learned GitHub workflow with feature branches
-Asana for organizing tasks
-Editing Zoom videos

Soft skills:

-Providing a team with directions, explaining technical ideas and teaching concepts
-Organizing meeting agendas, goal setting, and making sure the team is on the same page
-Ways to foster better communication and collaboration

Three Achievement Highlights:

  1. Demo to web scrape with Beautiful Soup and Selenium and provided team guidance

  2. Sample code to build a BERT model and use classification that gave a 0.80 accuracy

  3. Provided technical support for all members

Meetings and trainings attended:

Attended all team meetings
Attended or watched recordings of all industry training sessions

Next goals:
-Show team other ways to do classification
-Web app and cloud hosting ideas

Detail statement of tasks done:

  1. Week 1: Made a demo of web scraping in Beautiful soup on the MyPaint forum and taught it to the team, taught basic ML concepts, emphasized importance of data collection, gave detailed project overview and planned ahead (I have never web scraped before but I overcame this hurdle by using online resources and discussing with people who know how to)

  2. Week 2: Made a demo of web scraping with Selenium on the MyPaint forum and taught it to the team, provided technical support, designed dataset format details with feedback from team (I have never used Selenium before but overcame this hurdle with online resources and discussions)

  3. Week 3: Explained the data formatting, cleaning and processing to the team and introduced DistilBERT to get team started (I was not familiar with the input format of BERT so I was not sure how to format the data. I overcame this hurdle by discussing with others, attending the industry training sessions, and collecting more potentially useful data to be safe)

  4. Week 4: Sample code for DistilBERT and classification with 0.80 accuracy for additional support (I have never used BERT, PyTorch, or TensorFlow before. I overcame this hurdle by reading articles online, discussing with others, experimenting, and looking at the documentation and guides)

2

Self-Assessment 3

Concise overview of things learned:

Technical Areas:

-Web scraping a Discourse forum with Beautiful Soup and Selenium

-Learned and built BERT model with Hugging Face (PyTorch and TensorFlow)

-Learned to do classification with scikit-learn on last layer of BERT model output

-Learned about React and AWS for deploying the ML model

-Having a technical vision that is concrete yet flexible

-Reviewing code

Tools:

-Learned GitHub workflow with feature branches

-Asana for organizing tasks

-Editing Zoom videos

Soft skills:

-Providing a team with directions, explaining technical ideas and teaching concepts

-Organizing meeting agendas, goal setting, and making sure the team is on the same page

-Ways to foster better communication and collaboration

Three Achievement Highlights:

  1. Demo to web scrape with Beautiful Soup and Selenium and provided team guidance
  2. Sample code to build a BERT model and use classification that gave a 0.80 accuracy
  3. Provided technical support for all members

Meetings and trainings attended:

Attended all team meetings

Attended or watched recordings of all industry training sessions

Next goals:

-Improve model accuracy

-Introduce dynamic data

-Build on web app and cloud hosting

Detail statement of tasks done:

  1. Week 1: Made a demo of web scraping in Beautiful soup on the MyPaint forum and taught it to the team, taught basic ML concepts, emphasized importance of data collection, gave detailed project overview and planned ahead (I have never web scraped before but I overcame this hurdle by using online resources and discussing with people who know how to)
  2. Week 2: Made a demo of web scraping with Selenium on the MyPaint forum and taught it to the team, provided technical support, designed dataset format details with feedback from team (I have never used Selenium before but overcame this hurdle with online resources and discussions)
  3. Week 3: Explained the data formatting, cleaning and processing to the team and introduced DistilBERT to get team started (I was not familiar with the input format of BERT so I was not sure how to format the data. I overcame this hurdle by discussing with others, attending the industry training sessions, and collecting more potentially useful data to be safe)
  4. Week 4: Created sample code for DistilBERT and classification with 0.80 accuracy to provide additional support for the team (I have never used BERT, PyTorch, or TensorFlow before. I overcame this hurdle by reading articles online, discussing with others, experimenting, and looking at the documentation and guides)
  5. Week 5: Went over additional types of classification techniques. Introduced building a web app with React and AWS hosting. Let the teams present their ML modeling work.