01
Scale Engineering
Your system worked at ten thousand. It is folding at ten million. We find the real bottleneck — usually not where you think — and we rebuild the hot path without stopping the business.
San Diego, California · Est. 2019
BLUE Sapphire
Veteran engineers out of Google, Apple, Amazon and Meta, working alongside junior engineers out of SDSU, USD and UCSD. You get the judgment and the horsepower, on the same problem.
exGoogle◆exApple◆
exAmazon◆exMeta◆
exNetflix◆exStripe◆
Most consultancies sell you a deck. We send a squad that has already built the thing you are trying to build, at the scale you are trying to reach, and we stay until it is running in production.
0
median years of engineering before they joined us
0
big-tech engineering orgs represented on the bench
0
local engineering schools we hire juniors out of
0
of clients extend past the first statement of work
01 — The Practice
Six things we are genuinely the best at.
We say no to everything else. That is the whole trick.
02
AI & Applied ML
Past the demo. Evaluation harnesses, retrieval that actually retrieves, inference budgets that survive contact with finance, and a product surface people return to.
03
Platform & Infrastructure
The boring layer that decides how fast every other team moves. Build pipelines, environments, observability, and a paved road your engineers choose voluntarily.
04
Product Engineering
Interfaces with the fit and finish of a first-party app. Design and engineering in one room, one backlog, one standard — shipped weekly, not quarterly.
05
Security & Trust
Threat modeling, identity, secrets, supply chain, and the audit you have been quietly dreading. Findings you can action, ranked by what would actually hurt.
06
Technical Diligence
Two weeks, one document. What the team really has, what it will cost to keep, and where the bodies are buried. Read by investors and boards before nine-figure decisions.
02 — The Bench
Both ends of the experience curve.
Squads mix veterans with junior engineers — recent graduates out of SDSU, USD and UCSD. Some have already made the expensive mistakes at Google or Apple scale and know exactly which corner not to cut. Others are deep in current tooling, moving fast, with something to prove. They work the same tickets.
It is the model that made all of us good in the first place, and it is why we hire out of the three engineering schools on our doorstep rather than shipping the work somewhere cheaper. The veterans teach, the graduates catch up fast, and you pay a blended rate rather than a partner rate for everything.
Veteran bench
01
11 yrs
Distributed Systems
Partner
A decade-plus inside hyperscale storage and serving infrastructure, including capacity planning for services carrying a billion daily users.
- Snowflake
02
14 yrs
Product Engineering
Partner
Years shipping on first-party consumer platforms, where the last five percent of polish is the product and review standards are unforgiving.
- Apple
- Airbnb
03
12 yrs
Applied ML
Principal
Ranking and recommendation systems at consumer scale, then the unglamorous work of making inference cheap enough to actually ship.
- Meta
- Nvidia
04
15 yrs
Platform
Principal
Build and release engineering in orgs where a slow pipeline costs thousands of engineer-hours a week. Deploys measured in minutes, not hours.
- Amazon
- Stripe
05
13 yrs
Security
Principal
Product security review, threat modelling and identity at companies where a breach is front-page news. Used to telling executives unwelcome things.
- Microsoft
- Netflix
06
16 yrs
Technical Diligence
Partner
Silicon and systems, then years of technical diligence on deals from Series B through acquisition. Writes for people who sign the cheque.
- Qualcomm
- Tesla
Junior bench · SDSU · USD · UCSD
01
3 yrs
Platform
Engineer
Three years in, owning deployment tooling end to end. Out of a university systems lab and yet to slow down.
SDSU graduate
02
2 yrs
Product
Engineer
Two years shipping design systems, and the reason a front-end review queue moves in hours instead of days.
USD graduate
03
4 yrs
Applied ML
Engineer
Four years on evaluation harnesses and data pipelines. Reads more papers in a week than most teams manage in a quarter.
UCSD graduate
03 — Engagements
Three ways to put veterans on the problem.
Embedded Squad
A self-sufficient unit inside your org — your repo, your standups, your review process. They own a surface end to end and hand it back documented, tested and instrumented.
- Veterans and junior engineers on the same squad
- Weekly demos against one agreed metric
- A named partner accountable throughout
Most requested
Diligence Sprint
Two weeks, one document. Architecture, team, security posture, delivery risk and what it will genuinely cost to keep the thing alive — written for people who sign the cheque.
- Read by investors and boards pre-decision
- Findings ranked by what would actually hurt
- Verbal readout inside 24 hours of close
Fractional Principal
A staff-plus engineer in the room for the decisions that are expensive to reverse. Architecture review, hiring loops, incident post-mortems, and the calls your team is too close to make.
- Standing weekly, plus escalation access
- Reviews your designs before they are built
- Rolls onto a squad if the work grows
04 — The Method
Four weeks to something real.
-
01
Week 1
Cut
A week inside your codebase, your dashboards and your standups. We come back with the actual problem — frequently not the one in the brief — and the shortest path through it.
-
02
Weeks 2–3
Facet
A narrow, load-bearing slice, built end to end and running in your environment. Not a prototype. The first production increment.
-
03
Weeks 4+
Polish
We drive the rest of the build alongside your engineers, in your repo, through your review process. Weekly demos against the metric we agreed on in week one.
-
04
Close
Set
Handover is the deliverable: runbooks, tests, dashboards and two weeks of pairing so it is genuinely yours. Then we leave. Good engineers make themselves unnecessary.
05 — Home
Built in San Diego.
We are eighteen miles from a wireless research corridor, a biotech cluster, a growing neurotech scene, a defense-systems industry and three good engineering schools — with rent that lets senior people stay senior instead of moving into management to afford a house.
That is not a lifestyle note. It is why our engineers are still engineers fifteen years in, and why they are still here.
- Studio
- Little Italy, San Diego, CA
- Hours
- Pacific Time · 8am – 6pm
- Travel
- On site with you when it matters
06 — Proof
The shape of the work.
Engagements are under NDA, so what follows is how the work actually goes rather than a wall of numbers you cannot check.
Scale Engineering
Rebuilding a hot path
A write path that worked at launch volume and folds at current volume. We find where the time actually goes — usually not where the team assumed — rebuild that path behind a flag, and cut over without downtime.
AI & Applied ML
Making a retrieval stack answer
Retrieval that returns confident answers from the wrong documents. We build the evaluation harness before touching the model, so “better” becomes a number the team can check rather than a feeling.
Platform
Getting the release train back
When the pipeline is the constraint on every roadmap conversation. We rebuild the build and the test topology, then hand back a paved road the client's own engineers extend without us.
Technical Diligence
Two weeks, one document
What the target actually has, what it costs to keep, and which risks are real. Written for the people who sign, with findings ranked by what would genuinely hurt rather than by how many there are.
07 — Engagements for Q3
Tell us the problem
nobody wants.
Send two paragraphs — what you are building and what is in the way. A partner reads every message and replies within two business days. If we are not the right people, we will tell you who is.
Blue Sapphire · Little Italy, San Diego, California