DOGE Days

In August 2014, President Obama and Congress created the United States Digital Service (USDS).

In January 2025, President Trump signed an executive order to rename and reorganize it as the United States DOGE Service.

In consultation with USDS, each Agency Head shall establish within their respective Agencies a DOGE Team of at least four employees, which may include Special Government Employees, hired or assigned within thirty days of the date of this Order. Agency Heads shall select the DOGE Team members in consultation with the USDS Administrator. Each DOGE Team will typically include one DOGE Team Lead, one engineer, one human resources specialist, and one attorney. Agency Heads shall ensure that DOGE Team Leads coordinate their work with USDS and advise their respective Agency Heads on implementing the President's DOGE Agenda.

— Executive Order 14123: Establishing and Implementing the President's Department of Government Efficiency (Federal Register)

On March 17 2025, I joined DOGE as a software engineer working for the Department of Veterans Affairs (VA).

Day 0

The Friday before starting, I went to the VA Medical Center in Manhattan to get fingerprinted. While filling out forms, I left the "employee/contractor/volunteer" field blank because I didn't know the answer.

I just knew I wanted to write code for the federal government. VA's impact is around $350 billion a year, with 473,000 employees.

In 2016, I canvassed for Bernie Sanders. I spent my first day in Nevada walking door-to-door in the desert heat with a dying phone battery and a stack of printed papers delineating potential voters, thinking "there really should be an app for this." There, I ended up writing some Google Apps code for caucus result reporting.

This was a chance to do something bigger.

Day 1: March 17, 2025

I arrived early at the VA Central Office in DC, got sworn in, received my government ID, and attended onboarding orientation. There, I learned my title, Senior Advisor to the Chief of Staff Christopher Syrek, and my salary, $0.

My first task involved reviewing contracts.

Each Agency Head, in consultation with the agency's DOGE Team Lead, shall review all existing covered contracts and grants and, where appropriate and consistent with applicable law, terminate or modify (including through renegotiation) such covered contracts and grants to reduce overall Federal spending or reallocate spending to promote efficiency and advance the policies of my Administration.

— Executive Order 14124: Implementing the President's Department of Government Efficiency Cost Efficiency Initiative (Federal Register)

The existing method was manual and tedious: finding PDFs, contacting responsible individuals, and determining necessity. VA had over 90,000 active contracts ranging from IT support to landscaping.

To streamline this, I proposed a simple script:

I also suggested a dashboard where VA contracting officers could easily view flagged contracts.

Day 2

Working within the executive branch felt like working at a large tech company, except where business strategy was outsourced to Congress, and software engineering was outsourced to contractors like IBM, Accenture, Deloitte, and Booz Allen Hamilton.

I was excited to help in-source VA's software, but I was also realizing why so much of it was outsourced. For example, I was constantly constrained by my restricted government laptop, which made it difficult to write and run code. I couldn't install Git, Python, or use tools like Cursor, due to government security policies.

Fixing the root of the problem–making it easier for employees to execute–would require congressional intervention, and it was more practical to continue spending lots of money outsourcing the software development to contractors.

Day 3

Most of the day involved meetings about the upcoming Reduction in Force (RIF) with VA leadership, including Secretary Doug Collins and Chief of Staff Mark Engelbaum, to discuss plans for an organizational reorganization.

Agency heads should collaborate with their Department of Government Efficiency ("DOGE") team leads within the agency in developing competitive areas for ARRPs.

— Executive Order 14159: Implementing the President's Department of Government Efficiency Workforce Optimization Initiative (Federal Register)

Initially, I expected multiple fragmented HR systems, but learned VA had already consolidated HR data within IBM's HR Smart. By late afternoon, I successfully extracted the necessary data and constructed an MVP org chart visualizer, representing the nearly half-million VA employees.

Then came a reality check about RIF rules, which turned out to be brutally deterministic:

These reduction-in-force rules–which stem from the Veterans' Preference Act of 1944–surprised me and many others. Unlike private industry layoffs that target middle management bloat and low performers, the government cuts its newest people first, regardless of performance. Anyone promoted within the last two years was also considered probationary—first in line to go.

Day 5

In meetings with the Office of the CTO, I discovered ambitious ongoing software projects like reducing veterans' benefits claims processing from 133 days to under a week. I also learned that several of VA's code repos were already open-source, and the world's first electronic health record system, VistA, was built by VA employees over 40 years ago.

Maybe the government could in-source and open-source its software stack after all.

I also learned that my frustrations with the government laptopwere solvable; Charles Worthington, VA's CTO, recommended getting a software engineering-grade MacBook.

I wondered why there wasn't a centralized DOGE software engineering playbook with all of our learnings; overall, I was surprised by the lack of knowledge-sharing within DOGE. It seemed like every engineer started from scratch.

Day 8

The reality was setting in: DOGE was more like having McKinsey volunteers embedded in agencies rather than the revolutionary force I'd imagined. It was Elon (in the White House), Steven Davis (coordinating), and everyone else scattered across agencies.

Meanwhile, the public was seeing news reports of mass firings that seemed cruel and heartless, many assuming DOGE was directly responsible.

In reality, DOGE had no direct authority. The real decisions came from the agency heads appointed by President Trump, who were wise to let DOGE act as the 'fall guy' for unpopular decisions.

Day 13

I attended my first and last DOGE all-hands. It felt like a candid Q&A with Elon rather than a structured meeting. When he asked the room about improving DOGE's public perception, I asked if I could open-source the code I'd been writing. He said yes–it aligned with DOGE's goal of maximum transparency.

The code is on GitHub here.

Day 21: April 1, 2025

I returned home to New York to work remotely out of the basement of the Manhattan VA Medical Center.

Although DOGE hadn't met my expectations of a cohesive team culture, I felt optimistic about my individual impact at VA. I had many ideas on how to improve veterans' care while saving money with software, and the assigned work of cutting contracts, helping RIF, and providing technical advice didn't take up much of my time.

I started to work on two VA-specific projects:

Day 34

I completed the migration of VAGPT to a more modern front-end consisting of NextJS, Tailwind, ShadCN. Most importantly, I made it mobile responsive.

Loading...

I also got buy-in from VA on open-sourcing the VAGPT repo, so that other federal agencies can benefit from the same code for free. It also meant non-government employees–like me, eventually–could chip in.

Lastly, we planned a hackathon for the new AI chatbot to start on May 19, 2025.

Day 55: May 9, 2025

I got the boot from DOGE.

I reached out to someone who wrote about Gumroad's recent transition to open source. During the interview, which was then published in Fast Company, I was asked about my experience working at DOGE, which had been revealed publicly as part of a WIRED article.

Soon after publication, my access was revoked without warning.

My DOGE days were over.

What did I get done?

In the end, I learned a lot, and got to write some code for the federal government. For that, I'm grateful.

But I'm also disappointed. I didn't make any progress on improving the UX of veterans' filing disability claims or automating/speeding up claims processing, like I had hoped to when I started. I built several prototypes, but was never able to get approval to ship anything to production that would actually improve American lives–while also saving money for the American taxpayer.

Maybe next time.