DISCLAIMER: All information contained related to Blue Origin is used as a basis of discussion and not actual data, or was pulled from publicly available sources.
JIMMY YIN
May 2026 · Interview
An interview presentation by

Jimmy
Yin.

Veteran·Engineer·MBA
Routine spaceflight is a personnel and operations challenge.
AudienceBlue Origin
DateMay 2026
FormatLong-form
Read time~12 min
Scroll
Jimmy Yin on R-Day at West Point
R-Day · West Point
Chapter 01 · The journey

The start of a 10-year journey.

From cadet to Army Officer. The same five threads still drive how I approach every team and every problem today.

01
Commander's intent.Start with the end goal in mind and use it as the guide. Everything nests within the end goal.
02
Shared purpose.Build teams that believe in the mission and each other.
03
Bias to build.Pave the road and build what's next.
Chapter 02 · How I work

Five threads that shape how I show up.

Each one shows up in how I approach problems, teams, and technology.

01
VETERAN

Mission first

Commander's intent, teamwork, unbreakable connections, shared purpose.

02
LEADER

Build the team

Intertwine personal and professional. People who believe in the mission and each other.

03
ENGINEER

Understand the how

Electrical Engineer foundation, Civil/Mech experience. Persistent curiosity about how things actually work.

04
EARLY ADOPTER

Push the edge

AI glasses, blockchain, VR headsets, LLMs. New tools become new processes.

05
BUSINESS THINKER

Connect to outcomes

MBA lens — finance, corporate governance, strategy. The why behind the what.

CrossFit competition
FITNESS
Hoku the corgi
HOKU
Chapter 03 · Outside the work
The things AI can't solve and shouldn't try to.

CrossFit L1 coach + competitor. Hoku: a reminder that some of the most rewarding parts of life are the ones you can't automate or optimize away.

Chapter 04 · Recent relevant project

Predicting Li-Ion battery State-of-Health.

UW Professional Masters Certificate in ML/DL — Final project, Fall 2025.

The why

From data center floor to battery prediction.

At my data center role, annual UPS maintenance required a battery discharge to verify functionality. Lifespan was treated as a fixed duration.

If we can predict SOH from a short discharge snippet, we can proactively replace the units showing accelerated decay and extend service life on healthy ones. The same idea translates directly to EVs, consumer electronics, anything with a durable battery on a maintenance budget.

DatasetNASA PCoE Li-Ion (2007)
ModelsLinear · MLP · GRU-RNN
GoalBattery SOH Prediction
01BASELINE

Linear Regression

The floor. Establishes what a non-learning model can extract from the snippet — anything ML must beat this to justify itself.

Sets the bar
02DEEP NN

MLP — Multilayer Perceptron

The simplest deep architecture. Tests whether a basic deep model can recognize patterns in the time series at all.

Captures average, loses slope
03RECURRENT

GRU-RNN

Sequence-aware model designed to surface hidden temporal patterns in the discharge curve.

Tracks decay direction better
Results

Promising. Not yet production-ready.

Both models track the trend. Neither is tight enough to act on a single battery's reading.
MLP predictions on the held-out test set
MLP — Test set predictions
GRU-RNN predictions on the held-out test set
GRU-RNN — Test set predictions

Lab vs. real world

Controlled environments hit single-digit % error. Field-noisy data is the open problem.

Hybrid LSTM

Stacking recurrent layers with classical features is the current frontier for SOH.

Temporal CNNs

TCNs are emerging as a competitor to RNNs for fixed-length signal windows.

Chapter 05 · The Blue Origin question

Doubling the New Glenn launch rate by 2029.

"If Blue Origin were to double its launch rate over the next three years, how would I build the operations strategy to get there?"

FROM ~12/yr → 24+/yr BY 2029
5.1 · Baseline assumptions

Anchoring the math first.

Target for 2026
~12/yr
One launch every 1–2 months. Booster reuse cadence ≈ 30 days (per AST Space Mobile); 2nd-stage production ≈ 12/yr (Ars Technica).
New target by 2029
24+/yr
At least one launch every two weeks. Doubles 2026 cadence.
Four buckets to hit the target — Business Dev, Manufacturing & Supply Chain, Operations & Infra, and Finance. For the next ~10 minutes I'll assume Biz Dev and Finance have signed off, and dive into the two internal levers.
5.2 · Four-bucket framework

Two market-facing. Two operational.

All four have to clear before cadence doubles. The two highlighted are where I'll spend the rest of the deck.

01
EXTERNAL

Business Dev / Strategy

Is there market demand at this cadence? Who has the capital and operational appetite to actually buy the launches?

Assumed approved
02
INTERNAL

Manufacturing & Supply Chain

Can we build New Glenn at this rate without starving external customers (e.g. BE-4 to ULA)?

Deep dive
03
INTERNAL

Operations & Infrastructure

Mission control, launch infra, recovery infra, support teams — what breaks first as cadence rises?

Deep dive
04
EXTERNAL

Finance

Capital to expand the bottlenecks the internal review surfaces. How do we de-risk CapEx — customer agreements, govt incentives?

Assumed approved
Bucket 02 · Manufacturing & Supply Chain

Walk the 0 → 1 lifecycle. Find what breaks first.

Sourcing, production, refurb, payload. Each pillar has its own throughput ceiling, and at 2× cadence, every one of them needs a deliberate answer.

1

Raw materials

  • Sole-source suppliers — single points of failure
  • Can vendors ramp without de-prioritizing other customers?
  • Long-lead items — identify and pre-buy
2

Production capacity

  • FL · AL · WA — engines, boosters, GS-2, fuel tanks, PLF
  • Floor space — consolidate or add a line?
  • Critical-path bottleneck: booster, 2nd stage, or engine?
  • Does 2029 force a new facility?
3

Refurb & inspection

  • Booster turnaround — what's the limiter?
  • If inspection finds an anomaly, what happens to retrieved boosters??
  • FIFO vs. Refurbish in parallel
4

Payload processing

  • PPF capacity vs. 24+ /yr cadence
  • Condense timelines or restructure sequence
  • Testing & integration as a downstream constraint
Bucket 03 · Operations & Infrastructure

What gets stressed first as cadence rises?

Address them in order of most-limiting.

Launch infrastructure

Where the pad and the recovery vehicle saturate.

LC-36 is approved for 12 launches/yr — is the cap regulatory or design? Both answers point to a 2nd LC with redundant or shared infra.
  • LOX / LNG / LH₂ supply rate and onsite storage — does the propellant logistics tail keep up at 2× cadence?
  • Transporter / Erector — what's the O&M cycle? Where's the unplanned downtime?
  • Jacklyn — can it sustain operations every ~15 days? If not, what's the limiter — vessel, crew, or maintenance window?
  • Maintenance windows + FAA launch windows — collapse the slack we currently rely on.
Personnel teams

Cadence is a people problem first.

Are we staffed for back-to-back ops?

Find the single (or few) points of contact across mission control, recovery, and integration.

Build redundancy and a deliberate hiring plan before the schedule forces it.

5.3 · Risks worth designing for now

At 24+/yr, single-point failures become program-level events.

Each one of these has a recent precedent in the industry. The mitigations belong in the plan today, not after the first loss.

Pad

Catastrophic launch failure at LC-36

Damage to LC-36 or critical test infrastructure. Reference: SpaceX Massey site (S36) — 10-month repair timeline.

Recovery

Booster landing failure on Jacklyn

A single recovery vessel is a single point of failure for any reusable booster cadence.

Weather

Florida hurricane exposure

Concentrated FL footprint (LC-36, manufacturing, integration) means one season can wipe out a quarter.

Range

Range / regulatory bottlenecks

FAA windows, range conflicts, neighbor launches.

B/2-9 IN, Korea, 2018
B/29BEB · Korea, 2018
Thank you
Q&A.
Routine spaceflight is, at its core, a personnel and operations challenge.
Jimmy Yin
Veteran · Engineer · MBA