5 ms·

> lots of opportunity to work on many different things. What could you use help with? If a software engineer could invest 2 weeks in producing something useful

> lots of opportunity to work on many different things. What could you use help with? If a software engineer could invest 2 weeks in producing something useful for your company (including adding a feature to an existing open source project), what would be a list of potential quick projects?

layoric 7y ago
Not a list but data analysis tooling is probably a large category of things that need work on. A lot of our customers are deep knowledge experts but not highly computer literate, a lot are good enough with Excel but still struggle with UTC conversion. Large amounts of effort goes into validating the quality of our data by our customers and we look for ways to help but it is a large domain to itself while we are still focusing on making our data better and more accessible. We’re slowly building trust by giving data freely to researchers but validation of our data, handling formats, date time conversions, carefully equating temporally and spatially, etc is still largely manual and significant barrier of entry for some. I’ll keep in mind to build a specific list at some point.
- mariushn 7y ago
  Thanks! Sidenote, feedback on https://solcast.com/rooftop-solar/free-pv-system-performance-estimation-tool/ https://solcast.com/rooftop-solar/free-pv-system-performance... from a non-specialist: I know the tool targets existing solar users, but as a potential user, it would help me answer the question "Should I get a 5kw solar panel?". For the same input on the left, I'd like to see historic monthly data for the past 3 years, telling me how much energy my potential panels would generate per year. Also, would be useful to include some info on how much of that energy would actually used when passing through batteries and all the system. If some panels 100kwh in a month, that doesn't mean I can simply subtract 100kwh from my energy bill, but how much could I subtract?
  - layoric 7y ago
    Ha, good eye for what I work on! Appreciate the feedback and likely we'll have more tooling is this space coming out over the next 12-24 months, and the data for others to build these tools sooner than that. We need to get the data correct and easily accessible first and then the hope is we won't be the only ones building the tools on top and others like yourself can see gaps in what is useful and map that to data we have. Specifically regarding "historic monthly data for the past 3 years..", you can tackle it multiple ways, but likely the most scalable for us is monthly averages of a "typical" year for a specific location rather than last X years we can look back on 10+ years to get a better idea "typical" and synthesize it down to monthly. Combining cost comparison, batteries and individual energy consumption is another level of complexity again and not something that can give accurate answers to without knowing highly specific details. How much you could subtract you could take a guess at by looking at your electricity bills and having a guess that >50% of your consumption is probably during the day when the sun is out, but again a couple of heavy usage days when it is cloudy will throw off your averages. There are lots of websites about the economics of getting solar/battery, so while where I work will produce good data to help with those calculations, it will always be individual behavior that determines about worth while the purchase is. As a rule of thumb (in Australia), it is around a 7 year payback at the moment, and batteries will not save you money at this stage.
Fordec 7y ago
I'm working on environmental software at the moment, the best unanswered service? A good data API for climate emergencies and disasters classification/push notifications. We can pull in rain data, satellite imagery etc. but correlating it with problems that have a lagging effect (eg. rain today means flooding next week) is a problem I find isn't super scalable from an entry level at the moment. Disaster response is mostly an even by event analysis today. It's not a hard problem, it's just tedious.
- layoric 7y ago
  That is a great area to work in! I use to work in a government agency in on a very similar area (bush fire hot spot tracking), so your efforts are appreciated! Tedious is one way to put it, but I agree the tech side of push notifications, servers, APIs etc isn't hard but as you've alluded to, disaster detection/classification is tricky and if you are interested in the space, government jobs tend to be the most stable as commercially it doesn't have an obvious "market" consumers are willing to pay for. Similar to utility scale sector in energy, you'll have a hard time as a small team convincing government and commercial weather operators that your detection or forecasting of events from satellite imagery etc are more accurate, so my advice would be to do this with your eyes open and produce as much data validation evidence as you can to share widely so others can validate it. Working with live satellite imagery has a lot of challenges, and pulling clean data out of them consistently. Each satellite is different and has its own biases, operational issues that can be tricky but working on these problems I think is a really useful skill so not time wasted IMO. Best of luck!

layoric 7y ago

Not a list but data analysis tooling is probably a large category of things that need work on. A lot of our customers are deep knowledge experts but not highly computer literate, a lot are good enough with Excel but still struggle with UTC conversion. Large amounts of effort goes into validating the quality of our data by our customers and we look for ways to help but it is a large domain to itself while we are still focusing on making our data better and more accessible. We’re slowly building trust by giving data freely to researchers but validation of our data, handling formats, date time conversions, carefully equating temporally and spatially, etc is still largely manual and significant barrier of entry for some. I’ll keep in mind to build a specific list at some point.

mariushn 7y ago
Thanks! Sidenote, feedback on https://solcast.com/rooftop-solar/free-pv-system-performance-estimation-tool/ https://solcast.com/rooftop-solar/free-pv-system-performance... from a non-specialist: I know the tool targets existing solar users, but as a potential user, it would help me answer the question "Should I get a 5kw solar panel?". For the same input on the left, I'd like to see historic monthly data for the past 3 years, telling me how much energy my potential panels would generate per year. Also, would be useful to include some info on how much of that energy would actually used when passing through batteries and all the system. If some panels 100kwh in a month, that doesn't mean I can simply subtract 100kwh from my energy bill, but how much could I subtract?
- layoric 7y ago
  Ha, good eye for what I work on! Appreciate the feedback and likely we'll have more tooling is this space coming out over the next 12-24 months, and the data for others to build these tools sooner than that. We need to get the data correct and easily accessible first and then the hope is we won't be the only ones building the tools on top and others like yourself can see gaps in what is useful and map that to data we have. Specifically regarding "historic monthly data for the past 3 years..", you can tackle it multiple ways, but likely the most scalable for us is monthly averages of a "typical" year for a specific location rather than last X years we can look back on 10+ years to get a better idea "typical" and synthesize it down to monthly. Combining cost comparison, batteries and individual energy consumption is another level of complexity again and not something that can give accurate answers to without knowing highly specific details. How much you could subtract you could take a guess at by looking at your electricity bills and having a guess that >50% of your consumption is probably during the day when the sun is out, but again a couple of heavy usage days when it is cloudy will throw off your averages. There are lots of websites about the economics of getting solar/battery, so while where I work will produce good data to help with those calculations, it will always be individual behavior that determines about worth while the purchase is. As a rule of thumb (in Australia), it is around a 7 year payback at the moment, and batteries will not save you money at this stage.

Fordec 7y ago

I'm working on environmental software at the moment, the best unanswered service? A good data API for climate emergencies and disasters classification/push notifications. We can pull in rain data, satellite imagery etc. but correlating it with problems that have a lagging effect (eg. rain today means flooding next week) is a problem I find isn't super scalable from an entry level at the moment. Disaster response is mostly an even by event analysis today. It's not a hard problem, it's just tedious.

layoric 7y ago
That is a great area to work in! I use to work in a government agency in on a very similar area (bush fire hot spot tracking), so your efforts are appreciated! Tedious is one way to put it, but I agree the tech side of push notifications, servers, APIs etc isn't hard but as you've alluded to, disaster detection/classification is tricky and if you are interested in the space, government jobs tend to be the most stable as commercially it doesn't have an obvious "market" consumers are willing to pay for. Similar to utility scale sector in energy, you'll have a hard time as a small team convincing government and commercial weather operators that your detection or forecasting of events from satellite imagery etc are more accurate, so my advice would be to do this with your eyes open and produce as much data validation evidence as you can to share widely so others can validate it. Working with live satellite imagery has a lot of challenges, and pulling clean data out of them consistently. Each satellite is different and has its own biases, operational issues that can be tricky but working on these problems I think is a really useful skill so not time wasted IMO. Best of luck!