đź“Ł Lead a Team, Build Your Portfolio, and Get Paid: Apply Today for June & July Sessions

🔆AIVIA Learn & Earn STEM-Away Virtual-Internships
1

Ready to take your AI skills to the next level?
We’re looking for Team Leads for the upcoming Virtual-Internship Sessions in June & July. You’ll guide a small team in recreating, extending, and presenting cutting-edge research in one of five exciting AI areas (Recommender & Decision Systems, Language Models & Generative AI, Computer Vision & Multimodal AI, AI in HealthCare & Medicine, or Domain-Specific AI).

Once lead selection is complete, we’ll share the list of available virtual-internships. We anticipate running 1-3 teams per area.


How It Works

  • Pick a Paper
    • Choose a research paper of interest to you in one of the five categories, or check out our recommended list. We’ll review the paper for suitability within the program structure.
    • You will work with your team to recreate some of the paper’s results and add your own contribution!
  • 3-Phase Project Structure
    • Phase 1: Research & Planning – Dive into the methodology, define your extension, and set milestones.
    • Phase 2: Implementation – Reproduce the paper’s results (partially or fully) and add an extension (e.g., new features, deeper analysis, etc.).
    • Phase 3: Presentation – Showcase your work in a final presentation, perfect for your portfolio!
    • Each phase includes a set of objectives that’s flexible yet structured enough to provide guidance and enable evaluations.
  • Flexibility & Support
    • Meet with your team as needed. One mandatory meeting per week with STEM-Away mentors for progress check-ins and Q&A.
    • 3-week total timeline to keep things fast-paced (yet still manageable).


Why Join as a Lead?

  • Build a Killer Portfolio: Lead a project that highlights both your technical prowess and leadership skills.
  • Get Paid: Receive $100/week (paid weekly upon successful completion of each check-in). Plus, an additional $50 award per week goes to the team member with the strongest contribution. We’re also in discussions with company sponsors who may offer additional financial incentives in the future.
  • Flexible: Choose the paper that excites you, set your team’s pace, and focus on what you love.
  • Stand Out: Gain leadership and research experience that makes you a top candidate for future internships, jobs, or grad school.


What We’re Looking For

  • Expert-Level Evaluation in your chosen category, demonstrated by achieving “Expert” or “Intermediate” in at least one AIVIA project within that category.
  • Proactive Mindset – ready to lead a small team, solve challenges, and deliver results.
  • Passion for AI – excited to dive deep into research and push the boundaries of what’s possible.
  • Selection Process: We may invite shortlisted candidates to a brief (15-minute) call to discuss leadership approach and project ideas before making final decisions. If multiple “Expert” applicants choose the same domain or paper, we’ll select leads based on project alignment, extension ideas, and leadership potential.


How to Apply

  • Click on @stemaway
  • Select Message in the user card that appears.
  • Use the Application Template below (no exceptions). Applications that don’t follow this template will not be considered.
  • Deadline: Rolling
  • Spaces for Leads are limited—secure your spot and get ready to build something awesome!

We can’t wait to see what you create!

2

Application Template

Subject: Lead Application

Body:

  • Chosen AI Category
  • Relevant AIVIA Expert Evaluation
  • Proposed Paper or Topic
  • Brief Pitch (2–3 lines)
  • STEM Level
  • Availability


Application Details

  • Chosen AI Category: Please choose one of the following: Recommender & Decision Systems, Language Models & Generative AI, Computer Vision & Multimodal AI, AI in Healthcare & Medicine, or Domain-Specific AI (other than Healthcare & Medicine).
  • Relevant AIVIA “Expert” Evaluation:
    • Which project did you earn your “Expert” rating in? Please share the project URL.
    • Note: The project must be from your chosen category. Selecting a more complex project may improve your chances, but a strong evaluation on any relevant project is sufficient.
    • You may skip this question if you selected Domain-Specific AI (other than Healthcare & Medicine). We’ll create a relevant project in that domain and send you the details.
  • Proposed Paper or Topic: If you already have a specific research paper in mind, list it here. Otherwise, pick from our suggestions below (check back for new ones!)
  • Brief Pitch: (2-3 lines): Share any initial ideas on how you’d extend or analyze the research.
  • STEM Level: Indicate your current academic or professional status (e.g., Undergraduate Junior, Undergraduate Senior, Master’s Student, PhD Student, Junior Professional, etc.).
  • Availability:
    • Pick a starting Monday in June or July. If flexible, say something like “any Monday,” “any Monday in June,” etc.
    • Please note that you are committing to a 3-week timeline with one mandatory meeting per week.
3

LLM-Rec: Personalized Recommendation via Prompting Large Language Models

Link: arXiv:2307.15780v3

Why It’s Good

  • Cutting-Edge: Leverages large language models (LLMs) for personalized recommendations, a rapidly growing area in AI research.
  • Code Availability: Code is not publicly available. Paper can be used as guide only.
  • Versatility: The paper explores multiple approaches to integrating LLMs into recommender systems, providing ample opportunities for experimentation and innovation.

Potential Extensions

  • Domain Adaptation: Apply or fine-tune the approach on a different dataset (e.g., MovieLens, e-commerce data).
  • Prompt Engineering: Explore how different prompts or few-shot techniques impact recommendation quality.



Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer

Link: arXiv:1910.10683

Why It’s Good:

  • Cutting-Edge: Introduces the Text-To-Text Transfer Transformer (T5), a unified framework for solving diverse NLP tasks like summarization, question answering, and classification. It’s a cornerstone in modern NLP research.
  • Code Availability: Code, datasets, and pre-trained models are publicly available on GitHub.
  • Versatility: The text-to-text format allows T5 to be applied to a wide range of tasks, making it a powerful tool for experimentation.

Potential Extensions:

  • Task Adaptation: Fine-tune T5 for new tasks like sentiment analysis or domain-specific text processing.
  • Prompt Engineering: Experiment with prompts to improve zero-shot or few-shot learning performance.
  • Efficiency Improvements: Explore techniques like model distillation to make T5 more efficient for real-world applications.