ATS NER Competition Instructions

Welcome to the ATS (Applicant Tracking System) NER competition! Participants will form teams of two and follow the four phases below.

---

📋 Setup

1. Form Teams: Two participants per team.

2. Dataset: Download the resume dataset from Kaggle: https://www.kaggle.com/datasets/gauravduttakiit/resume-dataset

3. Doccano Project:

   • Create a new project in Doccano with Sequence Labeling.
   • Import the provided labels.json (8 entity tags) via Labels → Import.

---

🚀 Phase 1: Annotation (30 minutes)

• Objective: Split the dataset, label as many resumes as possible, and export your annotations.

• Steps:

  1. In your Doccano project, upload half of the dataset for your team.

  2. Annotate the resumes using these entity tags:

     [
       {"text":"PERSON_NAME"},
       {"text":"EMAIL"},
       {"text":"PHONE"},
       {"text":"DEGREE"},
       {"text":"UNIVERSITY"},
       {"text":"JOB_TITLE"},
       {"text":"COMPANY"},
       {"text":"SKILL"}
     ]

  3. Export your annotated data in JSONL format when time is up.

⏱ Time limit: 0.5 hour

---

🤖 Phase 2: Model Training & Hub Upload (30 minutes)

• Objective: Fine-tune a spaCy NER model on your labeled data and publish it.

• Steps:

  1. Open Google Colab and install spaCy: !pip install spacy
  2. Convert your JSONL annotations into spaCy training format.
  3. Train a basic NER pipeline on your exported data.
  4. Save the trained model as a package and upload it to the Hugging Face Hub under your team name.

⏱ Time limit: 30 minutes

---

🖥️ Phase 3: UI, Docker & Deployment (45 minutes)

• Objective: Build a simple web UI for your NER service, containerize it, and deploy.

• Steps:

  1. Create a minimal web interface wit Streamlit that accepts resume text and highlights named entities.
  2. Write a Dockerfile to containerize your app and your spaCy model.
  3. Deploy your Docker container to Hamravesh.

⏱ Time limit: 45 minutes

---

💬 Phase 4: Community Engagement (15 minutes)

• Objective: Showcase your application and give feedback to peers.

• Steps:

  1. Post a brief demo of your deployed application (screenshots or link) in the competition forum.
  2. Comment on at least two other teams’ posts with constructive feedback.

⏱ Time limit: 15 minutes

---

Good luck to all teams—may the best ATS NER demo win!