Deep / Head of Engineering
LA · ex-NASA · Epsilon3 Hiring
▸ Vehicle Assembly · Live
Pad Ready
15+ Years · Mission Control → Epsilon3 Scroll · Building ↓
▸ T+ 02:14:36 · Apogee

Orbit achieved.

Mission continuing · Brief follows
Head of Engineering · Epsilon3

Deep.

Six years at NASA writing software for Mission Control. Software engineer to principal engineer leading the team. Code that flight controllers ran live, with astronauts on the other end.

Joined Epsilon3 as principal engineer to build the prototype. The prototype became the heart of the product, and the product became the #1 procedures platform in the world — software for rockets reaching orbit, landers headed to the Moon, satellites overhead, and the factory floors that build them.

Five years from prototype to Head of Engineering. The work is executive now: design the org, set technical strategy, make the build-vs-buy calls, stand in front of customers and investors. The proof is commercial — FedRAMP High opened the federal market, a database re-platformed live under flying customers, and a 25-engineer org where the leads were grown, not bought.

NASA · 2015—2021 Epsilon3 · 2021— Los Angeles · 25 engineers
01
01 Origin

Apprenticed at Mission Control.

Software engineer to principal engineer leading the team. I mentored, owned architecture, and worked directly with flight controllers — building the systems that ran the missions: telemetry, command, anomaly tooling. Code that had to work the first time, because the people on the other end were astronauts.

Era 2015 — 2021
Organization NASA
Trajectory SWE → Principal
Focus Real-time ground systems
02
02 Trajectory

From prototype to #1 in the world.

2021
Joined as principal engineer. Built the prototype.
IC
Heart of the product
2022
Prototype shipped. First customers flew missions on it.
Production launch
Hiring engineers
2023
Promoted to director. Engineers and designers under one roof.
Cross-functional
Shipped faster
2024
#1 procedures platform in the world.
Aerospace · Defense
Manufacturing scaling
2025
Head of Engineering. Designed the org, wrote the playbook, hired senior engineers.
Four squads · org playbook
FedRAMP High · live-DB migration
2026
Multiple modules going to market. The company scales.
25 engineers · hiring in LA
Manufacturing v2 in production
03
03 Operating numbers

By the numbers.

25engineers
Across four squads
4leads
Promoted from within
100%
Team retention, 2025
99.99%
Uptime — through a live re-platform
8×
Deploy velocity since 2021
FedRAMPhigh
Opened the federal market
04
04 Operating philosophy

Judgment, written down.

On AI
AI changed the economics of engineering, not the engineers.
What collapsed is the cost of exploration and the cost of acquiring context. I rebuilt how the team prototypes, tests, and onboards around that — and retired code-volume metrics, because they now measure the tool, not the engineer. We measure the system instead: team-level flow, recovery, reliability.
On compliance
Compliance is a moat, not a tax.
FedRAMP High was run as product strategy, not paperwork. It opened the federal market to the business and forced an architectural discipline — isolation, auditability, controlled change — that made the whole platform better. Competitors face 14 months of the same work to catch up.
On org design
Small squads, full ownership, no telephone game.
Four squads, each owning its modules end to end — architecture, quality, and the customer conversation. Engineers sit with the people who fly the missions. The org chart is a product decision: every layer between an engineer and a customer is latency.
On leadership
Executives who stop reading code start guessing.
I still review architecture, still prototype, still know where the bodies are buried in the codebase. Technical strategy made from the org chart's top floor is astrology — it has to be made from the ground, with the altitude to act on it.
05
05 The role

The job, end to end.

Five functions, one owner. The job is carrying all of them at once — and staying close enough to the code to be useful.

Technical strategy
Every architecture decision is a market decision. We spent 14 months on FedRAMP High when it looked like overhead — it opened the federal market. We re-platformed the database live, under customers flying missions, with zero downtime. Strategy here means making calls like those early, and answering for them either way.
FedRAMP High
Case study →
Hands-on engineering
I still write code that ships. The prototype that became the product. A 3D viewer rendering spacecraft CAD in the browser. An offline iOS SDK built for places with no signal and no second chances. When the team hits the hardest problem of the quarter, I want to be in the diff — that's where the real information is.
Prototype → product
Org & talent
A 25-engineer org built to work without me: four squads owning their modules end to end, leads grown from within, a leveling framework engineers trust because they helped shape it. Good org design is quiet — things ship, people stay, and nobody waits on one person.
4 squads
Leads grown within
Commercial engineering
Engineering carries commercial weight here. I take the customer call, run the investor review on real delivery data, and we turned a compliance program into something that helps deals close instead of slowing them down. Engineering shouldn't only be a cost center — done right, it's part of why customers sign.
Federal market
Investor reviews
The operating system
Judgment, codified: a large-feature playbook with blast-radius mapping, team-level DORA telemetry instead of vanity metrics, a handbook that turns hard-won decisions into defaults. The test I hold myself to: if a process only works when I'm in the room, it isn't finished.
Playbooks · DORA
Handbook
06
06 Right now

Currently in flight.

Agentic AI
Procedures that run themselves.
Agentic workflows inside the platform — AI that drafts procedures, checks preconditions, and advances steps, while operators supervise the mission instead of the paperwork.
Virtual reality
Execution beyond the screen.
Procedures in VR — train crews and walk the factory floor before the hardware exists, with every step rendered in the space where the work actually happens.
Performance
Super speed, everywhere.
A platform-wide speed program — every page, every query, every render. When a procedure stands between an operator and a launch window, fast isn't a feature. It's the product.
07
07 Contact

Let's talk.

The throughline of the work: taking technology from prototype to product to company advantage — and building the org that keeps doing it without me in the loop. If you're building something consequential, I'd like to hear about it.

Email deep@epsilon3.io ↗ LinkedIn in/deeptailor ↗ GitHub @deeptailor ↗ Resume /resume.pdf ↗