Science writer and former designer Steve McClure argues creativity is a learnable skill, like playing an instrument, developed through disciplined practices. He contrasts design training with PhD culture, emphasizing the need to separate idea generation from evaluation and judgment to avoid looping on standard ideas. McClure says constraints (like a picture frame or the structure of academic papers and methods) fuel creativity, while critique—common in design juries and echoed in peer review—requires detachment from one’s work and thick skin. For developing research questions, he recommends mind mapping, listing, and free writing, plus “writing down to write up” via annotated bibliographies to avoid endless literature rabbit holes. He notes AI can assist brainstorming and semantic associations but tends toward mainstream answers and must be guided by human interpretation and iterative writing.

00:00 Can Creativity Be Trained

01:13 Design School Thinking

03:29 Constraints Fuel Creativity

06:58 Critique And Thick Skin

09:28 Brainstorming Mistakes

10:25 From Mess To Question

12:23 Mind Mapping Method

14:19 Writing Down To Write Up

16:23 AI For Idea Generation

22:12 Peer Review And Paradigms

24:47 Simple Creativity Exercises

27:28 Originality And New Mediums

28:41 Closing Thanks

On the FundaYao podcast, Dr. Prosper Washaya (from Zimbabwe, now in the Czech Republic) discusses studying and working across Zimbabwe, China, and Europe and how cultural and language gaps shaped his experience, while noting academic similarities across continents. He describes moving from urban planning and GIS in Zimbabwe into remote sensing at Wuhan University, being surprised by Shanghai’s infrastructure and Wuhan’s multicultural foreign-student environment, and contrasting small-department supervision in Zimbabwe with larger student loads and access to visiting experts in China and the Czech Republic. He recounts industry work in Deqing, China, on InSAR deformation monitoring, flood mapping, vegetation monitoring with drones, and 3D reconstruction, emphasizing language challenges and later integration into Chinese society. He summarizes his Czech PhD on multi-source remote sensing of forest disturbances from the 2018 drought and bark beetle outbreak, highlights stronger interdisciplinary collaboration and PhD funding opportunities in the Czech Republic, compares data access and compute constraints, and advises prospective students to research programs and supervisor fit. He is now a postdoc focusing on foundation models and embeddings for remote sensing.

00:00 Welcome and Guest Intro

01:28 Remote Sensing Origin Story

02:55 Why Study in China

04:27 First Impressions in Wuhan

07:32 Teaching and Supervision Styles

11:42 Life as African Student

14:30 Industry Work in Deqing

19:38 PhD Research in Czechia

21:03 PhD Systems and Funding

28:58 Remote Sensing Data Access

34:51 Advice and Next Steps

38:23 Closing Thanks and Goodbye

Deep Learning for SAR: Beyond Hype, Toward Sensor-Aware Models and Practical Challenges

On the FundaYao Podcast, we interview Michael Schmitt (Professor of Earth Observation, University of the Armed Forces, Germany) and Ronny Hänsch (DLR Microwave and Radar Institute) about their book “Deep Learning for Synthetic Aperture Remote Sensing.” They argue deep learning for SAR is not overhyped, enabling capabilities beyond classic signal processing while complementing, not replacing, traditional methods. The book aims to cover the full SAR processing chain and bridge gaps between data scientists lacking SAR expertise and SAR experts lacking modern AI skills. They discuss the niche nature of deep learning for SAR, warn against treating SAR as simple grayscale images, and advocate sensor-aware modeling that respects SAR physics and geometry. They highlight labeling and compute bottlenecks, as well as the industry–academia compute gap.

00:00 Welcome and Guests

00:30 Is Deep Learning Overhyped

02:21 Why This Book Exists

03:20 Who the Book Is For

06:14 Editing and Writing Challenges

10:13 Why SAR Deep Learning Is Niche

11:41 Doing SAR Deep Learning Right

14:47 Labels and Data Bottlenecks

18:46 Sensor Aware Deep Learning

21:02 Compute Limits and Access

27:26 What Counts as Foundation Model

32:02 Skepticism and SAR Backbones

35:13 How to Learn Deep Learning

36:19 Closing Thanks