Suggestions
Laurent Picard
Tech lover
Laurent Picard is a Developer Relations Engineer at Google Cloud, based in Paris, France.12 He joined Google in 2017 and has been working there for about 7 years.2
Professional Background
Before joining Google, Laurent had an extensive career in the tech industry:
- He co-founded Bookeen, where he served as CTO for 20 years.12
- During his time at Bookeen, he pioneered the ebook industry and launched eight generations of Cybook ereaders.1
- He also developed a digital platform used by major retailers.1
Current Role and Expertise
At Google Cloud, Laurent focuses on:
- Exploring and sharing cloud technologies with fellow developers.1
- Demonstrating how cloud technologies can provide powerful capabilities to users.1
- Creating content related to Google Cloud Platform, including blog posts on topics such as video summarization, object tracking, and face detection.1
Personal Interests
Laurent is described as:
- A passionate developer interested in software, hardware, science, and future-shaping technologies.13
- A tech lover who enjoys exploring, learning, and sharing possibilities in the world of technology.35
Public Speaking and Content Creation
Laurent is active in the developer community:
- He speaks at conferences like DevCon and ML Conference, sharing his knowledge on Google Cloud Platform and AI technologies.46
- He creates content, including blog posts and presentations, on topics related to building smart solutions using machine learning without requiring deep expertise in the field.7
Laurent Picard's LinkedIn username is indeed "picardparis," as mentioned in the query.2
Highlights
🐍 Updated my notebook on example benefits of using Generative AI in education: ▶️ https://lnkd.in/eDbmNtjv 🚀 Switched models to Gemini 1.5 (from 1.0) …and learned a few things along the way.
💡 Learning #1 Both Gemini 1.5 Pro and Flash excel at education use cases. Here are their strengths: → Flash (smaller) brings speed and a lower cost → Pro (larger) manages more complex use cases
For a new problem, I'll try Flash first and, if needed, switch to Pro.
💡 Learning #2 Gemini 1.5 models can cover many inputs at once: → text, image, video, speech, document
For videos, frame & speech samples are analyzed with their timestamps, which gives you 3 possible sources of information: → Image → Text → Speech with the timestamp info!
🚀 This lets you do the following: → Find info that is seen, written, or heard → Consolidate partial info → Mention specific video locations with timestamps → Know when something happened
In the attached example, the single source of truth for what the device does is what's pronounced at 0'20''.
💡 Learning #3 Gemini 1.5 Flash was trained by Gemini 1.5 Pro through a process called "distillation", in which: → The larger (teacher) model transfers its most essential knowledge and skills to the smaller (student) model. → The smaller model can be seen as a compressed version of the larger model, optimized for speed/efficiency.
🐍 Had a blast at #EuroPython2024 this week 🤩 At our booth, had uncountable lovely discussions about Python, Cloud, Google… 🤯 Last break yesterday was crazy with 360 registered participants (out of 1,245 onsite attendees) and ??? people waiting for the results 🙃 Breakfast was completely in phase with me this morning: tired-happy 🙏 Big thanks to the EuroPython organizers (still running the w-e sprints!)
