The game changes considerably as the company gets larger. Gemini is the biggest company I've been at in a decade, and having a team this size means being able to ship massive amounts of diverse product with incredible speed.

It also grants the ability to specialize. At a 12-person company, the engineering team is doing it all: MVPs for customer research, hotfixes, finding security holes, troubleshooting container orchestration, and building customer-facing features. Having dedicated teams for SRE, QA, security, and so on means product engineers can focus on building robust, performant features. Which is a good thing, because it's possible to get away with things serving a few hundred customers that quickly fall over when serving tens of thousands.

The same goes internally. Four or five engineers can coordinate amongst themselves, hundreds cannot; the development process quickly devolves into chaos. Team dynamics enter into play, and organizational choices suddenly carry weight. Do teams have visibility into the planned features delivery? Is there a process for resolving disagreements? Are disparate teams re-inventing solutions to problems that have already been solved? Do team members feel they have a say, and are they comfortable calling out potential problems?

The importance of this sort of housekeeping meta-work is amplified with fully-remote teams, whether they're remote by choice or pushed into it by circumstance. Communication and 'working in public' become essential, and design documents or otherwise writing things down are no longer negotiable. Deciding how much process needs to be in place is a delicate balance, and supporting the well-being of an engineering team is as well.

Cryptocurrency, particularly bitcoin, is moving slowly into the realm of mainstream finance - which means high-stakes engineering decisions. If a cat picture repository disappears from the Internet for a couple of hours, customers will complain. If a bitcoin derivatives trading platform goes down for a couple of hours and people get trapped in positions, customers will ensure you see the inside of a courtroom.

This means saving us, as engineers, from ourselves. Find areas where humans are able to make costly mistakes and automate them (pipelines, automated testing). Anything that can't be automated, lock it down and share reponsibility (code reviews, runbooks for everything - including contingencies). And choose your core competencies *very* carefully and hire best-in-class third parties for the rest (custody of assets, network security, audits)

With a pivot into retail, the customer experience is still paramount. Crypto has gotten away with bad CX for years because the users tended to be technical and willing to put up with nonsense. The bar is low, but that's all the more reason to clear it. The frontend should be both usable and fast. Websockets are great, but constantly re-rendering entire component trees will negate a performant backend even on a desktop, let alone a 3 year old mobile device.

As a small team, optimizing this customer experience is your superpower. It hinges on forging bonds and trust internally, but you can move *incredibly* fast. I had great relationships with sales, support, and design - we could frequently have a feature or bug identified, triaged, designed, architected, developed, deployed, and a response back to the customer with in a day. Huge companies have immense resources and power, but can't do that. Use it to your advantage.

The writing is on the wall for fossil fuels. The cost per watt for solar panels has dropped more than 80% over the last decade, and big names like Tesla have pushed rooftop solar from the domain of hippies and preppers into the mainstream. We wanted to accelerate this transition to rewnewables by providing an incentive, independent of government subsidies or tax credits.

SolarCoin is a blockchain-based solar rewards program. Solar energy producers sign up to receive grants based on the kilowatt-hours they produce. We use blockchain technology to both provide an indisputable historical record of the grants issued as well as provide an open economy for the reward tokens.

Being at a startup means erasing "not my job" from your vocabulary. I wrote smart contracts and built tools to migrate to a new blockchain, built tools that cut our claim processing time from days to minutes, and other technical work. But I also wrote pitch decks, answered customer service tickets, managed our community, developed the long-term strategy, and managed and mentored a team to implement it.

We've incentivized 12 terawatt-hours of solar energy - enough to power a million US homes for a year - across 72 countries since 2014. We've joined IRENA, partnered with the UN, and are incorporated into the dashboard of the world's largest inverter manufacturer. And we're just getting started.

Makeshift's beat was the informal economy: prosthetics formed from milk jug plastic, Nigeria's booming movie industry, and bicycle-powered farm equipment. The stories focused on individuals and their creative solutions to problems, usually in resource-constrained environments, and every quarter a beautifully printed issue with 90+ pages of new ones showed up on subscribers' doorsteps.

Faced with resource constraints of our own, we added onto the core magazine product in as many directions as we could: one founder and I taught design classes at the School of Visual Arts and other venues and designed an award-winning course pack for UPenn, while other team members pursued video production and syndication partnerships – all without a physical office.

My reporting beat was a perennial favorite: high-tech crime and the shady side of the Internet. I've posted up several of my articles below.

Sadly, the economic realities of an indie print magazine caught up to Makeshift, but the content lives on at mkshft.org.

Before crowdsourcing became a household word (depending on your particular house), there was Mutopo. Beyond the spec-work competitions normally associated with the term, we were looking at something deeper and weirder: what does it mean to actually collaborate with people who don't work for your organization?

Along with the array of miscellaneous tasks and general hacking required to keep the wheels turning on a 3-person startup, I got to work on some awesome projects: community managing an open design competition to replace the paper coffee cup and one to cram a huge loft's functionality into a tiny apartment, putting together a conference on the future of work, and researching how people work together and why they contribute to a big open design project (spoiler alert: it isn't money).

Mutopo no longer lives in the US, although it continued for several more years in Brazil.