Sign In

Hannes Hapke

Machine Learning at Digits

Hannes Hapke is a seasoned professional in the field of machine learning and software engineering, with a strong focus on NLP and deep learning. He has co-authored publications like O'Reilly's 'Building ML Pipelines with TensorFlow' and Manning's 'NLP in Action'. Hannes has extensive experience in developing deep learning systems and machine learning pipelines, particularly working with TensorFlow Extended and Kubeflow Pipelines. His M.Sc. thesis delved into Biomimetic Control Strategies with applications of Neural Networks and Genetic Algorithms. Apart from his machine learning prowess, Hannes is proficient in Python programming and Django, with deep knowledge of REST API development and various Python ecosystem tools. He has managed commercial projects, contributed to open-source initiatives, and co-founded a technology startup called renooble, Inc. Hannes has developed applications across diverse verticals such as healthcare, recruiting, commercial real estate, renewable energy, energy markets, and patient management in healthcare. With a background in Electrical Engineering from Oregon State University and the University of Stuttgart, Hannes has held various roles at reputable organizations like Google, SAP Concur, and Manning Publications Co.

Highlights

Feb 18 · linkedin

Google's release of #gemini 1.5 pretty stunning! At Digits, we got early access to Gemini 1.5, and we were blown away by the initial results we saw.

Seeing models with a 1M token context window is incredible. What a breakthrough!

We tested the Vision model, and it was able to determine the location of all images we tested it with. The level of explanations was great. The model explained a tax question is detail, or provided the steps to solve a math problem (also the math was correct).

We also liked the safety aspects built into the UI!

Another snippet that got overlooked in the initial announcement: The model was trained with significant less data than Gemini 1.0, but the results were on par with Gemini Ultra. That is very encouraging for next generation open source models.

Thank you, #Google and #deepmind, for the early access.

Hannes Hapke
Add to my network

Location

Portland, Oregon, United States