Menu Mentor Hiring Testimonials Research Bootcamp Apply now BLAST AI An intensive online AI research program for talented high school students. 2025 applications are currently being accepted. Apply Now Explore the exciting and novel projects completed by BLAST students in past cohorts! Read Student Publications Developed by researchers from We've guided students from 250+ high schools Our alumni continue their journey at prestigious institutions including: Where curious students become accomplished researchers Our mission is to inspire life-long learning BLAST AI is a launchpad for aspiring research scientists and innovators alike. Our students don't learn to copy pre-written code, they discover new architectures and present at peer-reviewed conferences. Through an instructional program designed by researchers and PhDs from Berkeley and Stanford, students delve into the depths of machine learning, exploring everything from classical algorithms to cutting-edge techniques. If you are curious and passionate about AI, we'd love to welcome you to our next cohort! *BLAST AI students Annie and Zhou En at the 2023 ICTC Conference Research Highlights Explore the latest research projects from our students. AuPPLE: Augmenting Physical Priors Fine-tuning physical priors into large language models. GANaxy: Galaxy Image Anomaly Detection Using contrastive learning to achieve state of the art accuracies in galaxy anomaly detection. SCALE: Semantic Code Analysis via Learned Embeddings Achieving code isomorphism via contrastive learning and parameter-efficient fine-tuning. Image to Image Diffusion For Underwater Image Enhancement Designing an image-to-image pipeline with CLIP for efficient domain transfer in latent diffusion models. Skills from Large Language Model-guided Agents Yield Success Using agented LLMs and document retrieval to simulate an reinforcement learning. Deep Learning Approaches for Optimized Web Accessibility Correcting DOM violations and enhancing user experience with LLM design analysis. Unsupervised Learning of Molecular Embeddings Enhanced clustering and property detection in compounds using tanimoto molecular fingerprinting and contrastive learning. AuPPLE: Augmenting Physical Priors Fine-tuning physical priors into large language models. GANaxy: Galaxy Image Anomaly Detection Using contrastive learning to achieve state of the art accuracies in galaxy anomaly detection. SCALE: Semantic Code Analysis via Learned Embeddings Achieving code isomorphism via contrastive learning and parameter-efficient fine-tuning. Image to Image Diffusion For Underwater Image Enhancement Designing an image-to-image pipeline with CLIP for efficient domain transfer in latent diffusion models. Skills from Large Language Model-guided Agents Yield Success Using agented LLMs and document retrieval to simulate an reinforcement learning. Deep Learning Approaches for Optimized Web Accessibility Correcting DOM violations and enhancing user experience with LLM design analysis. Unsupervised Learning of Molecular Embeddings Enhanced clustering and property detection in compounds using tanimoto molecular fingerprinting and contrastive learning. AuPPLE: Augmenting Physical Priors Fine-tuning physical priors into large language models. GANaxy: Galaxy Image Anomaly Detection Using contrastive learning to achieve state of the art accuracies in galaxy anomaly detection. SCALE: Semantic Code Analysis via Learned Embeddings Achieving code isomorphism via contrastive learning and parameter-efficient fine-tuning. Image to Image Diffusion For Underwater Image Enhancement Designing an image-to-image pipeline with CLIP for efficient domain transfer in latent diffusion models. Program Timeline Throughout the intensive 8-week program, you will learn the fundamentals of machine learning and conduct a novel research project with an experienced mentor. Week 1 Bootcamp Week 1: Classical Machine Learning Week 2 Bootcamp Week 2: Deep Learning Week 3 Research Week 1: Literature Review Week 4 Research Week 2: Infrastructure Setup and Write Introduction Section Week 5 Research Week 3: Initial Experiments and Write Methodology Section Week 6 Research Week 4: Continue Experiments and Write Results Section Week 7 Research Week 5: Final Experiments and Conclusion Section Week 8 Research Week 6: Refine Paper and Prepare for Symposium Presentation Our approach to learning BLAST programs are designed to be exhilarating! With hands-on learning and individualized mentorship, our students design deep learning architectures for real-life applications. Hands-on learning philosophy Our students learn from real-world challenges BLAST AI brings AI to life with exciting projects. Through carefully crafted exercises, students develop a keen intuition for classical machine learning theory. Students master deep learning architecture design through collaborative experimentation, gaining invaluable practical experience. BLAST AI teaches students to combine creativity, teamwork, and innovation through hands-on learning, readying them for tomorrow's AI challenges. Our commitment to results BLAST students work on cutting-edge projects BLAST AI students aren't just learning; they're innovating, with nearly 20 projects accepted or awaiting approval at PhD-level conferences. Our students have placed in the top 1% of Kaggle competitions, set new benchmarks for AI DOM accessibility, and even surpassing OpenAI embedding models in code alignment. With BLAST AI, you're not just studying AI, you're shaping the future of the field. Visit the research page to learn more! Our focus on community BLAST students form a tight-knit network BLAST AI is more than just an instructional program; it's a vibrant community hub. Our interactive workshops and fun community events transform online learning into a dynamic, social experience. Here, students form lasting connections, sharing insights and sparking ideas that lead to collaboration well beyond the program. Join BLAST AI and be part of a community that thrives on creativity, support, and the collective pursuit of life-long learning. Mentor Spotlight BLAST Mentors come from all different backgrounds - research, medicine, and industry. Each mentor brings a unique skillset that allows BLAST students to tackle difficult, ambitious problems. Peter Wu Peter is a PhD candidate at Berkeley researching low-resource NLP and AI for healthcare. He is published in NeurIPS and Nature. Amisha Kumar Amisha is currently attending med school. She's conducted genomics research with Nobel Laureates at Stanford and Harvard. Rishi Jain Rishi will attend grad school for ML this fall. He researches RL for task generalization, parameter-efficient tuning of LLMs, and speech science. Our Global Student Community BLAST AI students come from all around the world. Frequently asked questions When are applications due? When will we find out if we were accepted?Research Applications for 2025 will be accepted in two rounds. Students who apply during the early round (due February 28, 2025 at 11:59 PM PST) will be given priority. Regular round applications will remain open until April 30, 2025. Both rounds will be decided on a rolling basis How much does BLAST cost? Are scholarships or financial aid available?The 8-week BLAST AI Research Program fees will be $1460.Financial aid is available for both programs based on need. Historically, 30% of students attend on financial aid and we sincerely hope the fees do not stop you from attending. BLAST is committed to making AI research accessible, which is why we charge less than a fifth of the price of comparable programs. Can international students apply?Absolutely! Students from 20+ countries have attended BLAST AI in the past, and we value a diverse set of backgrounds. For reference, meetings / lectures begin at 8 am PST to accommodate most timezones. When is BLAST?BLAST Research will run June 17, 2025 to Aug 12, 2025. I'd love to apply and join BLAST. However, I don't have much experience in AI or computer science.We're looking for motivated, passionate, and talented students regardless of whether you have experience. BLAST will be rigorous, but if you think you can keep up with lessons and assignments, we definitely recommend applying. Over the duration of BLAST, can I participate in other activities?The bootcamp will include 4 hours of instruction each day, and you should allot 2-3 additional hours per day for homework, projects, and office hours. The research program will have structured lectures once a week, and mentors will meet with groups 2-3 times per week throughout the 8 weeks. 2025 Summer Applications Are Now Open! Apply Now Contact Curriculum FAQ's © 2025 BLAST AI, all rights reserved