We bridge the technology gap through hands-on experiential learning. From empowering educators at under-resourced institutions to building real AI capability within enterprises, we prepare people to build AI, not only use it.
By equipping faculty and students who carry build-AI practice into classrooms and careers, we aim to scale toward impacting hundreds of thousands of students.
AFI partners with under-resourced institutions, including HBCUs, TCUs, MSIs, and community colleges, to deliver practical skills in AI, quantum technologies, and emerging innovations.
We empower faculty and staff to build AI models and agents themselves, so they can lead their students through the coming technological shift with confidence and creativity. Student pathways build the same hands-on literacy for the AI-shaped workplace.
Scale goal
By training faculty who bring build-AI practice into their courses, and students who carry skills into the workforce, we aim to scale toward impacting hundreds of thousands of students.
Partner with us
We welcome host campuses, consortium partners, and grant collaborators in workforce development, AI education, and broadening participation.
Separate cohorts for instructors who will teach what they build, and for students building project experience. Same hands-on DNA, different goals.
Programs serve every academic discipline, ensuring AI readiness across the curriculum, not only computer science.
Flexible delivery based on community needs, informed by multi-year virtual and in-person HackHPC programs.
Flagship experience
Our scalable hands-on intensive: fine-tune open models, learn the pipeline, and take build-AI practice back to the classroom or career.
1-3 day sessions for senior leaders to identify high-value use cases and develop internal AI roadmaps.
Intensive project-based work moving cross-functional teams from concept to deployed AI agents on proprietary processes.
Training selected employees to lead ongoing AI development, ensuring long-term institutional knowledge.
The same principle guides our enterprise work: organizations need people who can build AI on their own processes and data, not only experiment with consumer tools. Hands-on programs put teams into real execution inside secure, company-controlled environments.
Code, collaborate, and fine-tune. Not your average training course. The Megathon is a high-energy hybrid event that takes participants from a theoretical baseline to a deployed fine-tuning project in focused sprints, with the intimate culture of a small codeathon and room for 100+ people.
Small, efficient open models are changing what teams can run without a massive cloud bill. In a Megathon, your mission is to take a base open foundation model and fine-tune it on a specialized dataset until it behaves like a domain expert, small enough for local inference on ordinary hardware.
You will not do this alone. Teams use prompt-augmented learning: AI coding assistants help write training loops and structure data, while instructors explain the what and why so you learn the pipeline, not only the shortcuts.
Events adapt for faculty or students as separate cohorts, and run in person or virtually. We partner especially with under-resourced institutions, including HBCUs, TCUs, MSIs, and community colleges.
Scale goal
Distributed mentorship and parallel judging let us serve 100+ attendees without the classic "wait for the instructor" bottleneck. Faculty who bring this practice into courses multiply impact across classrooms, toward hundreds of thousands of students.
Learn it in sprints. Teach it when you go home. The Megathon is designed so participants leave ready to repeat the pipeline, not only watch a demo.
Balanced pods of 4-5 people, clustered by complementary skills. Four teams form a Group with a dedicated Moderator so support stays close as the room grows.
Just-in-time micro-lectures (about 10-15 minutes) between coding phases teach the next milestone. Then teams execute for 45-60 minutes with prompt-augmented support.
Teams pitch inside their Group first. Winners advance to a plenary stage for the full crowd. Less waiting, more energy, more learning time.
Traditional codeathons stall when everyone queues for a handful of mentors. The Megathon uses a tiered support model so main instructors orchestrate learning instead of debugging every line.
Deliver just-in-time micro-lectures between sprints covering technical, theoretical, and applied concepts. They keep the whole event on track rather than acting as the only debug queue.
Host-institution moderators, pre-trained with a runbook, oversee about four teams. They facilitate collaboration, unblock architecture-level issues, and keep groups moving.
Each team of 4-5 has a lead who coordinates the project and represents the pod. Balanced skill mix (for example data work and communication) is intentional, not accidental.
Within a Group, teams compete and still share solutions: collaborative competition that keeps the feel of a small group while the room scales to hundreds of people.
Challenges stack in sequence so every team ships something real: data in, better model out, results you can show.
Clean and structure domain-specific data (for example QA pairs in formats ready for training).
Load a small open model, run it on your task, and record how it fails before fine-tuning.
Set up parameter-efficient fine-tuning (PEFT / LoRA) with prompt-augmented coding for the training loop.
Test the fine-tuned model, show the performance delta, and present a simple interface or results story.
Prompt-augmented engineering
Use AI coding assistants to write, debug, and understand advanced Python and training code, with instructors explaining the generated pipeline.
Data curation for LLMs
Prep raw material into training-ready datasets that actually change model behavior.
Parameter-efficient fine-tuning
The mechanics of LoRA and related methods for small language models you can run locally.
Applied teamwork under pressure
Real agile collaboration in a high-energy environment: useful for classrooms, research groups, and the workplace.
Host or partner
We welcome campuses, consortia, and grant collaborators ready to host a Megathon or train moderators. Reach out to explore faculty or student cohorts for your community.
Our academic work and the Megathon scale a mentor-supported, project-based model refined across student codeathons and faculty-focused events since 2020.
Workforce Events
Student and faculty programs through the HackHPC initiative since 2020.
Megathon Scale
Designed for large cohorts without sacrificing high-touch team support.
Hundreds of Thousands
Students reached over time as faculty builders multiply impact in their classrooms.
Browse past student and faculty event materials, teams, and outcomes on the HackHPC event archive.
Visit hackhpc.github.ioAbundant Frontier Institute is a Washington State Nonprofit Corporation (UBI: 606 212 555) and a federally recognized 501(c)(3) tax-exempt organization. Our mission is to democratize access to advanced technological knowledge and empower a tech-enabled public workforce.
Donations support academic programs, Megathons, faculty and student pathways, and partnerships with under-resourced institutions. Contributions are tax-deductible to the extent allowed by law. Contact us for gift details and acknowledgment.
Explore academic partnerships, host a Megathon, or talk corporate capability programs. We want to hear from you.
connect@abundantfrontier.org