Ry Walker and
the trouble with arriving

Ry Walker had acquired a title with a peculiar problem. It sounded like an ending. After co-founding Astronomer and helping it through its early years, he was a board member and founder emeritus. In a 2025 conversation with Nathan Latka, he recalled feeling too young for the honorary-sounding designation.

There was affection in the account, and a clear assessment of where he could be useful. Astronomer had a leadership team. Walker had already hired another CEO. The company could keep growing without requiring him to occupy the chair at the center of every decision. He had helped build an organization capable of outgrowing his preferred job.

That job was starting things. Leaving in 2022, after selling shares in a secondary transaction, gave him the chance to do it again. A secondary sale lets an existing shareholder sell stock; the company need not be acquired. For Walker, an asset on paper became money he could actually use. The next company would be Tembo. The next business model would prove less settled.

“I'm too young to have a title that has the word Emeritus in it.”

Ry Walker · conversation with Nathan Latka, 2025

Before the databases, there were websites

Walker's career reaches back to a considerably less automated internet. His personal chronology records computer science studies at the University of Cincinnati from 1990 to 1992, followed by leaving without a degree. It also records a development shop that grew to 30 people. He later held technology leadership roles through the years of Ruby on Rails.

The continuity is easy to miss if the story starts with Astronomer. He had already spent years making software and working inside other businesses before co-founding Differential in 2013. The venture studio worked with Meteor, a JavaScript development framework. It offered a way to make products with other founders, with design and engineering close together.

A studio also presents a temptation to a product-minded founder: every promising client problem can look like the company you ought to be running. Walker wanted to commit to one. In an early interview, he described exchanging his Differential equity for a larger stake in USERcycle, the analytics project that became the starting point for Astronomer. It was a consequential way to narrow his attention.

A company learns what it is for

Astronomer began in 2015. Its eventual identity as an Airflow business was reached through changes, rather than delivered intact at incorporation. The early work involved clickstream analytics: following the trail of activity people leave as they use software. Over time, the team moved toward the infrastructure that carries data between systems.

A January 2017 interview offers a small, dated window into the business before its later scale. Walker reported a team of 25 and 20 customers. The value of those figures is their modesty. The company was still a group of people trying to persuade a relatively short list of customers that a difficult piece of infrastructure deserved a budget.

Apache Airflow provided a more specific focus. It lets teams define and schedule workflows in code, making dependent jobs run in the proper order. Walker has described bringing its creator, Maxime Beauchemin, into Astronomer's orbit as an adviser. Knowing which technology to build around mattered; knowing the person who had created it mattered too.

The later result can obscure all the intermediate versions. Walker's own account includes uncomfortable decisions about whether to continue with the original idea. A pivot asks a team to surrender work it can see for an opportunity it cannot yet prove. It is a fairly expensive form of changing one's mind, even when it turns out well.

The same city, another starting line

Cincinnati remains a fixed point in a career full of revisions. In February 2024, the University of Cincinnati announced that Tembo and Fireroad would share space at its 1819 Innovation Hub. The pairing brought Walker together with Tim Metzner, his Differential co-founder, in the city's innovation district. Their companies had changed; the relationship continued.

The university setting also gives the story a pleasing loop. Walker had left a computer science course decades earlier. Now a company he founded was putting down roots in a university innovation hub. There is no need to turn that into a sermon about degrees. It is simply a reminder that a relationship with a place can last longer than a student registration.

Tembo's first proposition concerned Postgres, the open-source database. The business aimed to make its capabilities easier for developers and companies to use. In July 2024, Tembo announced a $14 million Series A led by GreatPoint Ventures, with participation from investors including Venrock, Grand Ventures and Wireframe Ventures. That financing belonged to the database chapter.

The announcement described a team adding Postgres specialists, including David Wheeler, creator of the PostgreSQL Extension Network. This was a substantive investment in a particular future. A funding round can make that future appear fixed: a number, a lead investor, a plan. What it cannot provide is a guarantee that the plan will remain the most persuasive use of the company.

The metric that required a human yes

In May 2025, Tembo changed direction toward an AI developer teammate. By his July 30 interview that year, Walker was talking about a different unit of progress: merged pull requests. A pull request proposes changes to a software project. Merging it means the changes have been accepted into the codebase. It is a useful place to ask whether generated work has become useful work.

Walker reported more than 1,000 merged pull requests and 30 paying customers at that point. These are a snapshot of the early product, rather than a statement about its size today. What makes the choice of metric interesting is the approval embedded in it. Producing code is one event. Getting that code accepted by the people responsible for the project is another.

Tembo now describes a platform connecting coding agents to repositories, team tools and approvals. Its founding team includes CTO Darren Baldwin Jr., CPO Benjamin Akar and CGO Connor Paton alongside Walker. The company is trying to make agents useful inside the working arrangements of an engineering team. Even software that can write software needs somewhere to put its work.

A dashboard customers did not want

Another Walker project supplies a less flattering, more revealing example of how he changes course. OSSRank attempted to evaluate open-source projects using activity, contributors and commercial support. The intuition was sensible: a popular repository can still be a risky dependency. A star count alone says little about whether anyone will maintain the thing you have just built into your business.

In May 2026, Walker wrote about shutting OSSRank down. People found the analysis useful, he explained, but did not want an ongoing subscription. They wanted an answer about a few projects at the moment a decision had to be made. The customer was asking for a report while the company was offering continuing access to software.

It is the sort of distinction that looks obvious after someone has paid to discover it. Walker accepted that the delivery format had been wrong and kept the evaluation checklist. The closure belongs in the same story as the better-known companies. It shows a willingness to put an unflattering result on the record, rather than quietly letting a project disappear from the biography.

Four jobs, fewer comfortable assumptions

His management advice has a similar appetite for movement. In June 2026, Walker argued for giving a promising young hire four different jobs over two years. He favors breadth early: let someone work across functions, learn where demand actually sits, and become capable of more than the first job description imagined.

There is a practical condition. Before moving up, the person helps recruit a replacement for the work being handed off. An aspiring salesperson does not simply abandon the prospecting job. The next person has to arrive. This makes advancement a small exercise in company-building, complete with the inconvenience of making sure somebody else can do the work.

He also makes room for intuition, provided it can be challenged. His January 2026 essay on the subject treats experienced judgment as accumulated pattern recognition. The obligation is to test the hunch cheaply. That qualification matters for someone whose career contains several pivots. A founder can change direction quickly and still owe colleagues evidence that the new direction deserves their time.

Still willing to explain the idea

Away from the company descriptions, Walker lists golf, fishing, nature, science and art among his interests. His public GitHub profile still identifies him as a coder alongside founder and investor. There is something grounding about that ordering of activities: the business can change its product, its customers and its pitch while the person remains interested in making things work.

His March 2026 essay about arriving early dwells on the period before a market recognizes a category. He describes the repetitive labor of explaining why something should matter, keeping spending low and staying close to the few customers who already have the problem. It is a less cinematic account of entrepreneurship than the ceremonial photograph with a funding number.

Walker now wants Tembo to provide the controls that let companies adopt coding agents: visibility, choice of tools and human review. His manifesto puts the ambition plainly: “Trust the model; control the environment.” It is a claim about how work should be organized, and it remains a business proposition to be tested. After websites, a venture studio, data workflows and a database company, he is again asking people to try a different way of working. Emeritus will have to wait.