Enrico Shippole
Chief Executive Officer at Teraflop AI
Boca Raton, Florida, United States
Summary
👤
Senior🎓
Top SchoolEnrico Shippole is a CEO and ML engineer with nine years of experience specializing in large language models, transformers, NLP, and computer vision, now leading Teraflop AI from Boca Raton. He combines hands-on research and production work—contributing open-source adaptations like an AliBi-based Pathways variant and enhancements to PaLM and Toolformer repositories—to bridge cutting-edge model architectures with practical tool integrations (calculator, wiki search, calendar, and flash attention on A100 GPUs). Known for integrating parallel transformer blocks and PyTorch 2.0 flash attention, he focuses on performance and real-world utility rather than purely theoretical advances. Trained in economics at the University of Miami, he brings a product-minded, cross-disciplinary perspective to ML engineering and leadership.
9 years of coding experience
Economics College of Arts and Sciences, Economics College of Arts and Sciences at University of MiamiSt. Mark's School
English
Stackoverflow
77reputation
6kreached
2answers
10questions
Github Skills (20)
calculator10
transformers10
apim10
machine-translation10
pytorch10
artificial-intelligence10
python10
attention-mechanism10
reinforcement-learning10
flash10
deep-learning10
api10
self-attention10
multihead-attention10
cuda8
Programming languages (12)
Github contributions (5)
lucidrains/toolformer-pytorch
Feb 2023 - Feb 2023
Implementation of Toolformer, Language Models That Can Use Tools, by MetaAI
Role in this project:
ML Engineer
ML EngineerContributions:7 commits, 1 PR, 8 comments in 2 days
Contributions summary:Enrico implemented and integrated various tools within the `toolformer-pytorch` repository. These tools include a calculator, Wikipedia search, and machine translation capabilities, demonstrating an understanding of API integration and utility development. The user added a calendar tool and refined existing tool functionalities and prompts to align with the paper's approach, enhancing the model's ability to interact with external resources. Further contributions focused on model architecture, specifically adding a GPTJ transformer block and incorporating the use of parallel transformer blocks to the project.
api-callingartificial-intelligenceattention-mechanismsdeep-learningtransformers
lucidrains/PaLM-rlhf-pytorch
Mar 2023 - Mar 2023
Implementation of RLHF (Reinforcement Learning with Human Feedback) on top of the PaLM architecture. Basically ChatGPT but with PaLM
Role in this project:ML Engineer
ML EngineerContributions:6 reviews, 9 commits, 1 PR in 3 days
Contributions summary:Enrico focused on integrating PyTorch 2.0's flash attention into the PaLM model. Their commits introduce and modify the `palm.py` file, adding code to enable flash attention, including logic for checking PyTorch version, GPU compatibility (A100 specifically), and handling masks. They also address syntax errors, making updates to ensure only flash attention is enabled and expanding the attention mask.
transformerspalmdeep-learningreinforcement-learningfeedback
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.