👋 Johannes Hötter

Kern AI

At Kern AI, we're building the next-gen AI platform helping teams to collaborate and successfully implement LLM use cases.

Co-Founder & CEO

path2.ai

I founded path2.ai during my master's studies together with Henrik Wenck. We wanted to understand what problems companies face when implementing meaningful artificial intelligence. During this time, we helped organizations take their first steps towards AI, e.g. by implementing pilot projects or presenting AI in workshops. We also helped existing AI teams by providing insights into the state of research in computer vision, natural language processing, and document classification.

Founder & CEO

GAMBIT Consulting GmbH

Associate Data Migration Consultant

Hochschule Bonn-Rhein-Sieg

Tutor Programming and Linear Algebra

My main goal at Kern AI is to help enterprise teams successfully deploy Large Language Model (LLM) applications in their daily business. 

To do so, I post regularly about AI best practices in theory, strategy, and hands-on use cases - and at Kern AI, we build the collaborative next-gen AI platform to put everything into practice.

At Kern AI (previously onetask), we aim to build a dev environment for AI training data with which we aim to provide both higher-quality and higher-quantity training data. If you are interested in how we do this, check out our website www.kern.ai

There's never been a better time to check out this resource. It is my #1 book on AI, all-time. It addresses so well-written how jobs/roles change if certain tasks become extremely efficient to be solved - which is what we're seeing e.g. in software engineering or medical use cases. Perfectly describes the scenario we currently are in regarding AI.

https://lnkd.in/eZGmdHFG

Geschäftsführer Kern AI (acq. by accompio)

English content on LI: "We deployed AI agents and here is what happened"
German content on LI: Another rant about Deutsche Bahn

Come on ...

adesso SE

Head of Machine Learning & GenAI - Google Cloud

Senior Software Engineer - ML / Data - Google Cloud

ML / Data Engineer - Google Cloud

Junior ML / Data Engineer - Google Cloud

AI Guild

Member

myBuddy gUG

- Led the development and operation of the myBuddy online platform
- Coordinated meetings with internal and external stakeholders
- Facilitated the establishment of 500+ intercultural friendships through the myBuddy friendship program
- Assisted the founder with various organizational tasks

Platform Manager

Fraunhofer Institute for Production Technology IPT

Master Thesis: "Benchmarking of Deep Learning Algorithms for Stem Cell Classification"

- Implemented a modular pipeline for biomedical image processing in the AIxCell research project
- Applied deep learning and automated machine learning for classification, segmentation, and quantification of mono- and binuclear cardiomyocytes in confocal microscopy imaging data
- Enabled assessment of proliferation in postnatal cardiomyocytes to examine cardiac plasticity
- Published a peer-reviewed paper, "A Modular Deep Learning Pipeline for Cell Culture Analysis: Investigating the Proliferation of Cardiomyocytes," at MIDL2022

Graduate Research Assistant - Master Thesis

TH Köln

- Collaborated on statistics and machine learning projects with a team of research assistants and professors
- Organized and managed the audit process for written and online exams
- Taught mathematics tutorials for probability and statistics to bachelor's degree students
- Conducted pre-study mathematics courses for new students
- Created and published YouTube tutorials on probability and statistics during the COVID-19 pandemic

Graduate Research And Teaching Assistant

Center for Advanced Life Cycle Engineering (CALCE)

Bachelor Thesis "Degradation Mechanisms of Aluminum Electrolytic Capacitors"

- Evaluated the efficiency of reformation processes for aluminum oxide dielectric layers

Undergraduate Research Assistant - Bachelor Thesis

Pepperl+Fuchs GmbH

Working Student

Daimler AG

Electronics development department, team energy management (EvoBus GmbH)

- Analyzed and identified patterns in global battery and generator failure data
- Collaborated on manual inspections and evaluations of failed parts
- Developed and implemented a VBA tool for data management and visualization
- Contributed to high-voltage energy management and tested functions in various environments in the field

Practical Internship Semester

wob AG

Project Management Social Media

We build Machine Learning and Generative AI solutions across various domains in the Google Cloud. 

Julius Haas

I started working with Peer Robotics. I will focus on Computer Vision, Edge Deployments, and Cloud Infrastructure. 

We enable collaborative robots to become Artificially Intelligent and learn like humans.

Checkout the Peer3000: 
https://lnkd.in/dWTJZHGd

Building AI Solutions | DMs open | M.Sc. Automation Engineering

If you’ve worked on a project in with AI or ML before, you know that having data in sufficient quality and quantity is essential. The data is often much more important than the model.

NVIDIA has open-sourced NeMo Data Designer, a framework for generating synthetic data from scratch or based on your own seed data.

It’s fully open source and comes with some great features. You can generate data using different strategies and plug in your own model endpoints, including locally running models.

I have tested it to generate some datasets and the output looks great. It is not rocket science, but takes away lot of manual steps and saves you time when you quickly want to generate a dataset to validate an idea.

That said, I have yet to see a system trained primarily on synthetic data perform consistently well over time in real-world scenarios. Data drift and feature drift will almost certainly catch up with you eventually. Getting the distribution right in your data is key, and that will require real-world data at some point.

  
Start here: https://lnkd.in/gnGz_Czv
Repo: https://lnkd.in/gk65uFWP
Documentation: https://lnkd.in/gR-zReYE

You haven’t truly tested your resilience until you’ve tried to fix NVIDIA/CUDA dependency issues during a Docker build for some hours. After spending most of the day doing exactly that, my smartwatch politely suggested I should rest for 66 hours. 😀

I agree, Garmin. I agree.

The Stanford ML group around Andrew Ng just released an agentic paper reviewer. 

Upload your paper, and get a review with suggestions on how to improve your work. 

With more and more peer-reviewed papers being generated using LLMs, we’ll basically now have LLMs reviewing what other LLMs have written.

Too bad they don’t expose this via an API or MCP server, we could fully automate the whole process end to end and finally get it over with.

My kind of humor… 😀

See for yourself: https://paperreview.ai/

taplio

The Battle

👋 Johannes Hötter

Julius Haas

Followers

Posts

Engagements

Likes

Performance Score👋 Johannes Hötter's LinkedIn Performance

Performance ScoreJulius Haas's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

👋 Johannes Hötter

Julius Haas

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}👋 Johannes Hötter's LinkedIn Performance

Performance ScoreJulius Haas's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance Score👋 Johannes Hötter's LinkedIn Performance