The data I have is here: https://github.com/radeeyate/StratoSpore/blob/main/software/... - just be warned that the altitude data still isn't the exact same as it was while in the air (GPS not working so I had to take it from someone else).
> UV light, a form of energy, is defined as light having wavelengths between 100 nanometers (nm, 1 billionth of a meter in length) and 400 nm. [...]
> Most acrylic plastics will allow light of wavelength greater than 375 nm to pass through the material, but they will not allow UV-C wavelengths (100–290 nm) to pass through.
In terms of photonic permittivity, Glass is better for cold frames and the like, because acrylic filters out UV light.
Also, Hydrogen peroxide (H2O2) is an algaecide.
/? hydrogen peroxide algaecide https://www.google.com/search?q=hydrogen+peroxide+algaecide
Meanwhile in my attempt with High altitude balloon, I tried sending a whole image over Lora successfully of course in chunks.
https://codetiger.github.io/blog/sending-large-data-like-ima...
Who cares, though? Scientists train for many years to learn the details of experimental methods in their specific domain. The engineering and hacking experience on its own is what really matters here.
What stands out most isn’t the biological hypothesis (which, as others mentioned, may need tighter controls and shielding to isolate variables), but the systems-level thinking required to make the entire experiment function end-to-end.
A few things you nailed that many researchers and engineers struggle with:
Design under constraints Turning a 1080p frame into an 18×10 LoRA-safe transmission is exactly the kind of creativity real-world engineering requires. Constraints build better thinking than unlimited compute.
Cross-disciplinary integration Hardware, software, RF, micro-climate biology, and ML rarely play nicely together. Getting them working in one pipeline — even imperfectly, is already a level-up.
Failure as signal, not setback Losing the physical payload but retaining telemetry is a feature, not a flaw. Space systems are designed under the same idea: assume recovery is unlikely, design for forward-only information flow.
If you iterate this again, it would be interesting to see:
A ground control dataset logged in parallel
Shielded vs unshielded algae samples
Active excitation instead of ambient-only measurement
A comparison model using classical regression vs ML
But even without a perfect biological signal, the engineering and thought process here is solid. Projects like this are how people enter aerospace, robotics, or defense research roles — because they demonstrate initiative, constraint-driven problem solving, and resilience.
Well done, keep going.
Not affiliated with Hacker News or Y Combinator. We simply enrich the public API with analytics.