Noteworthy Side Projects

Video Game Speedruns

While working on my game backlog, I tried speedrunning a few shorter video games (I'm a fairly competitive person).

Actually managed to get top-ranked in completion time for portions of Pacman Championship Edition 2 and a small bullet hell game. Just like logistics and engineering, breaking a large problem into smaller pieces makes it much easier to find improvements.

First, I got the basics down consistently with minimal errors (don't die and waste time restarting the level). This dropped completion time by about 10%.

Then, I looked for smaller but riskier optimizations around specific bottlenecks. This mainly revolved around stage bosses (the only portions of the game that can be completed faster based on player skill).

For example, one can position their mech in advance to hit bosses off-screen and avoid pre-programmed attacks to increase Damage per Second (DPS) and clear stages faster.

Also, one can exploit bugs to damage bosses before the actual fight begins during the start of some stages. This shaved off another few percentage points (the difference in getting 1st place and just placing).

I may never be good enough at speedrunning to raise money for well-respected charities like Awesome Games Done Quick (an old Custom Google Map with Houston Food Bank Charity Restaurants with a few thousand views is likely the best I can do on my own), but it was definitely a neat experience.

Multiple Website Shopping Program

During the start of the COVID crisis, retail supply chains broke spectacularly. That left me (and a lot of others) scrambling. With scenes of completely empty grocery aisles, tensions rose quickly. A Custom Google Map I'd made years ago with Houston hurricane supplies locations gave us an edge in finding hand sanitizer/toilet paper/alcohol wipes in-store. Most forgot about international grocers, dollar stores, and sports retailers and opted to wait in hours of lines at big box stores like Costco instead.

Despite that, It wasn't just my immediate family I was responsible for. I suddenly also had to help coordinate groceries and cleaning supplies for far away elderly extended family with no tech literacy or internet (who couldn't safely go into supermarkets with an unknown virus spreading).

Thankfully, the internet is a wide net. Even if prices were high, most goods could be found online with an exhaustive enough search (though lead times were longer).

Unfortunately, no one retailer had even 25% of the things my family needed. Even worse, high demand items like toilet paper and disinfectant could run out in MINUTES (sometimes in just a few seconds after midnight website updates).

To ensure access to critical supplies for my grandparents and other relatives, I wrote a python script to search a list of sites that had the highest chances of having what we needed (and google shopping which would help me discover smaller specialty sites like Netrition for my relatives with food intolerances).

Nothing elegant, of course, but in a high stress situation, Keep It Simple Stupid (KISS) is an EXTREMELY helpful mantra. File available for download here (requires Python to run).

This combined with weekly supplies checklists (modified from work) for each household kept errors to a minumum. Thankfully, everyone was kept comfortable until supply chains became more stable a few months in.

PowerBI Nutrition Dashboard

After a month or two of lockdown, I decided to try 3 new things: to cook more, to get fit (or at least as fit as I was while playing high school baseball), and to shave my head. Two of these three tests turned out well.

A few months of experiments later, my list of foods (50+) and macronutrients to track (fat, protein, fiber, potassium, etc) were getting bloated. I needed an efficient way to review my options to make good nutritional choices while ALSO not increasing my food budget. A long spreadsheet and intuition weren't optimal.

To handle the conflicting parameters of cost and nutrition, I made new personal Key Performance Indicators (KPIs). In this case, grams of nutrient per calorie and nutrient per $1 spent. This way, I could get a good understanding of inexpensive foods that ALSO fulfilled my dietary needs.

Calories per dollar spent was also an important metric for low calorie snacks. I couldn't afford to spend oodles of money to eat the nutritional equivalent of sawdust just to keep me from overeating.

The results had some surprises. I expected eggs to be a good value overall (high fat and protein for low cost). I did NOT expect popsicles to be more frugal as snacks than even bulk protein powder.

Used PowerBI visualization for professional development. The biggest datasets I've used for software testing and model optimization were only tens of thousands of rows. Nothing big enough to need my knowledge of SQL or other databases (and PowerBI's drag and drop systems made me prefer it over Tableau).

Auto-updating with new data and the memory-light measure feature are also neat.

Mobile access is also appreciated (Excel on my ancient cell phone quickly becomes visual soup).