Steven J. McClain — Manufacturing Engineering Leader

OP 05

The Short Version

Ten years of getting hardware out the door. I started out running a 200 person lab under FDA rules, spent six years in an aerospace machine shop doing everything from 5-axis programming to running the IT, and now I lead the engineering side of a shop building solid rocket motor nozzle components, V-22 hardware, and launch-vehicle structures.

Along the way I kept hitting the same wall: the software the floor needed didn't exist. So I built it — an MES, a machine-health scoring system, a full metrics program with its own validation rules. I'd rather build the tool than wait for one.

I like factories more than slide decks. I like schedules that hold because the system works. If that's the kind of operation you're running — or the kind you want — keep scrolling.

OP 10

The Rate Story

The hardest transition in hardware isn't design — it's the day a qualified product has to ship every month. That's the work I do now.

Job shop in, production line out

At Votaw Engineering I'm leading the conversion of a job shop into rate production for GEM 63 solid rocket motor nozzle components and V-22 flight hardware — developing and implementing the manufacturing methods and processes, reorganizing floor flow for throughput, and setting the standard work that holds a sustained ship rate of 2 nozzle sets per month.

Launch vehicle hardware, first-hand

I serve as responsible manufacturing engineer on new launch vehicle structural hardware — owning the process plan, fixturing strategy, and VTL / 5-axis programming from ERB kickoff through first article — for customers including Relativity Space, Lockheed Martin, Blue Origin, and Northrop Grumman.

Rate is a systems problem

Heroics don't scale; systems do. I built a custom MES from scratch (SQL Server / Python) that gives the floor live visibility, and I author the analytics and governance — delivery, yield, capacity, machine health — that tell us where the constraint is moving before the schedule does.

People make the rate

I have lead a five person team of manufacturing and design engineers plus IT — setting team and individual goals, driving them to completion, and partnering with design engineering, quality, and production on build issues, NCR dispositions, and corrective-action closure.

OP 15

Programs, Kickoff to Delivery

Machining doesn't ship hardware. Somebody owning the whole thread does — requirements, schedule, reviews, anomalies, and the customer call when something moves. On my programs, that somebody is me.

LIFECYCLE — KICKOFF → DELIVERY

Own it end to end

I run the kickoff — design review, material readiness, risk register, schedule baseline — then carry the hardware through build, first-article and readiness reviews, and close with the acceptance data package. One owner, no handoff gaps.

CUSTOMER INTERFACE

The technical voice across the table

Requirements interpretation, change negotiation, and status — for Northrop Grumman, Janicki, Lockheed Martin, Blue Origin, and Relativity Space.

ANOMALY RESOLUTION

When reality pushes back

NCR dispositions, root cause driven to closure, and the judgment call on when to stop the line. I'd rather lose a day than ship a questionable part.

FRAMEWORKS

No single points of failure

Documentation standards, cross training gap flags, and metrics with named owners — built so the program survives any individual, including me. That's the entire point of the intelligence program in OP 20.

OP 20

Systems I Build

Most shops run on tribal knowledge and spreadsheets. I replace both with governed, validated systems the whole floor can see. These are real programs I authored — shown at the methodology level, no customer data.

FEATURED — METRICS GOVERNANCE FRAMEWORK

Manufacturing Intelligence Program

A complete operating framework I wrote for running a plant on data: four north-star questions (Are we executing? Are machines working? Are we making good parts? Are we efficient?), a full metric catalog with tiered sigma validation targets — 6σ for safety, contract, and traceability data down to 3σ for trend signals — five validation methods borrowed from gage R&R thinking, automated decision signals for hiring, redeployment, and capital purchases, and an 8-layer roadmap from clean data to predictive ML. Written so the program survives any individual — including me.

North-star questions every metric must serve

Validation tiers, 6σ → 3σ by decision risk

Validation methods, cross-source to SPC-on-metrics

Layers from clean data to predictive analytics

MES — BUILT FROM SCRATCH

Live production visibility

A custom manufacturing execution system in SQL Server and Python: order status, station queues, and throughput on screens.

MACHINE HEALTH — FLEET-WIDE

7-input Machine Health Score

A weighted reliability / productivity / maintenance model with red gate logic, scored per machine across the full fleet and fed by daily loss driver data — setup, no work, no operator, down. One glance says which machine is bleeding hours, and why.

ANALYTICS — MAKE VS. BUY

Outsourced processing ROI

Python/SQL extraction of 300,000+ purchase order lines — roughly $2M of penetrant, chem-film, anodize, and etch spend — bundled by qualified process to build the business case for bringing capability in house.

OP 30

Leadership

From a five person engineering team to a 200 person regulated operation — the job is the same: define the standard, remove ambiguity, keep promises to the schedule.

FEB 2026 – PRESENT

Votaw Engineering

PROCESS ENGINEER — PRODUCTION SYSTEMS & THROUGHPUT LEAD

Leading a 5-person team (manufacturing/design engineers + IT) through a development to production transition on flight propulsion and launch-vehicle hardware. Responsible engineer from ERB kickoff through first article; built the MES, machine-health system, and Manufacturing Intelligence Program; project-manage contracted work from Relativity Space, Lockheed Martin, Blue Origin, and Northrop Grumman.

2020 – FEB 2026

Aero Precision Engineering Inc.

ENGINEERING MANAGER

Managed engineering for precision machined aerospace structural components across the Relativity Space, Northrop, Blue Origin, Lockheed, ULA, Boeing, Honeywell, and Janicki supply chains. Led a 6-person team of machinists and engineers, ran the company's entire IT function, programmed 4/5-axis machining, designed fixtures and flight hardware, and owned customer technical relations with Northrop Grumman and Janicki.

2016 – 2020

PMD Laboratory

OPERATIONS MANAGER

Directed a ~200 person operation in a regulated CLIA/FDA environment. Standardized procedures and operating metrics improved throughput 18% while holding full regulatory compliance — proof that discipline and rate are the same discipline.

OP 40

Hands On Depth

Leadership only works if you can stand at the machine and be right. Current program work first — described at the hardware class level, no customer part data — then the process library that shows how I think.

IN WORK — CURRENT PROGRAMS

TRAVELER 101VTL · 5-AXIS

SRM nozzle components — at rate

CLASS: SOLID ROCKET MOTOR NOZZLE · MODE: PRODUCTION

OP 10	TURN — ROUGH & FINISH	VTL
OP 20	CONTOUR FEATURES	5-AXIS
OP 30	DIMENSIONAL	CMM
OP 40	NDT / PROCESSING	VENDOR
OP 50	FINAL ACCEPT & SHIP	QC

I own the manufacturing methods behind the ship rate — routings, fixturing, work instructions, and floor flow holding 2 nozzle sets per month — plus the NCR and root-cause loop when reality pushes back.

STATUS · AT RATE

TRAVELER 102RESPONSIBLE ENGR

Launch-vehicle propellant-tank closure

CLASS: STAGE-1 TANK STRUCTURE · MODE: FIRST ARTICLE

OP 10	ERB / PROCESS PLAN	MFG ENG
OP 20	FIXTURE & PROGRAM	NX / CAM
OP 30	LARGE-DIA MACHINING	VTL · 5-AX
OP 40	IN-PROCESS VERIFY	PROBE · CMM
OP 50	ACCEPTANCE DATA PKG	QC

Responsible engineer on a first-article, large-diameter thin-wall tank closure for a new launch vehicle — owning the process plan, fixturing strategy, and programming from engineering review board through acceptance.

STATUS · FIRST ARTICLE IN WORK

TRAVELER 103NC PROGRAMMING

Nozzle inner wall — final machining

CLASS: LARGE NOZZLE STRUCTURE · MODE: PROGRAM KICKOFF

OP 10	KICKOFF / READINESS	PROGRAM
OP 20	PROCESS PLAN & MO	MFG ENG
OP 30	TURNING	VTL
OP 40	CONTOUR FEATURES	5-AXIS
OP 50	ADP / PACK & SHIP	QC

Manufacturing engineering and NC programming for final machining of a large nozzle inner wall — from program kickoff (design review, material readiness, risk register) through the acceptance data package.

STATUS · IN WORK

PLAYBOOK LIBRARY — HOW I RUN PROGRAMS

PB 001 · DEV → RATESee → Stabilize → Set Rate

My 90-day transition playbook: value-stream map and takt-vs-demand math first; standard work and constraint relief second; a rate-readiness plan staffed to takt third. Heroics are a bug, not a feature.

PB 002 · RATE MATHTakt, constraint & capacity modeling

Demand → required starts after yield → station math that names the constraint and prices every lever: stations, touch time, test capacity, FPY. Run it yourself in OP 50.

PB 003 · TEST THROUGHPUTDebottlenecking acceptance

Inspection queues are the constraint OEE never shows. I model inspection as a work center, parallelize capacity, and gate WIP so accepted hardware — not built hardware — sets the rate.

PB 004 · METRIC GOVERNANCEValidated metrics, tiered by risk

Every metric earns its place against four north-star questions, carries one source of truth, and is validated to a sigma tier matched to the decision it drives — 6σ for safety and contract data, 3σ for trends.

PB 005 · CAPITALThree reasons to buy a machine

Add capacity, replace aging, or buy capability — each with its own signal set and math: pre-buy OEE check, true annual cost of keeping, break-even utilization. Capital argued in dollars, not adjectives.

PB 006 · WORKFORCEHire, redeploy, or go get work

Decision signals that separate headcount problems from distribution problems from backlog problems — sustained overtime %, utilization, No-Staff hours by machine — before anyone approves a req.

HANDS-ON 007 · 5-AXISHigh-accuracy 5-axis workholding

Datum strategy, dovetail-then-soft-jaw fixturing, probing for automatic datum verification → repeatable tolerance control with fewer reclamps.

HANDS-ON 008 · CAMCycle-time cut on complex surfaces

Engagement analysis, tuned step-over and tilt, high-efficiency semi-finish → ~22–30% cycle-time reduction at full dimensional and cosmetic quality.

TECHNICAL DEPTH — MISSION HARDWARE

TECH 01 · GD&TDatum strategy for large thin-wall structures

Large-diameter, low-rigidity domes and rings distort under their own clamping force. I build the datum scheme off functional interfaces, not machinability, and validate with in-process probing before committing to finish passes — so the delivered part matches design intent, not just the fixture.

TECH 02 · CLEANLINESSPropellant-system contamination control

Oxidizer- and fuel-wetted hardware can't tolerate machining residue, cutting fluid, or particulate. I write precision-clean processes into the router itself — bag-and-tag points, cleanroom transitions, particulate limits — cleanliness is built into the sequence.

TECH 03 · JOININGWeld/braze qualification for pressure boundaries

Pressure-boundary joints don't get a second chance. I build qualification around the real service condition — proof pressure, leak rate, thermal cycling — and lock process parameters to a controlled procedure so every joint after qualification is the same joint.

TECH 04 · TESTLeak check & proof pressure planning

Every flight pressure vessel gets proved before it's trusted. I plan test sequence, media, and acceptance criteria around the failure mode that actually matters — external leak, internal leak, structural margin — instead of running every test at every station.

TECH 05 · FIXTURINGVacuum fixturing for large dome structures

A thin dome has no rigid place to hold onto — clamp it wrong and you machine the clamp's shape, not the part's. I've designed large vacuum fixtures for dome hardware and backed the hold with both hand calculations and computer analysis before cutting chips, so the fixture's holding force is proven, not assumed.

TECH 06 · MATERIALSCryogenic behavior in process planning

Metal doesn't behave at -300°F the way it does on the shop floor. Thermal contraction, embrittlement risk, and seal-material compatibility get built into the process plan itself.

Live program cards are described at the hardware class level only — no customer part numbers, drawings, or controlled data. Playbook entries summarize methods I run on real programs; hands on entries use generic demo geometry.

OP 50

Rate Readiness Calculator

The question every hardware program asks: can the floor actually hit the manifest? This demo runs the same math I use on real transitions — takt vs. touch time, test capacity, first pass yield — and names the constraint.

RATE READINESS — RESULTS

Shippable capacity (after yield)–units / mo

Starts required for target–units / mo

Constraint–

Gap to target–units / mo

Touch-time cut to hit rate as-is–

Enter your line parameters and run the check.

Demonstration only — simplified on purpose. Real rate readiness work also carries learning curves, WIP policy, staffing to takt, and supplier lead times. The point is the decomposition: name the constraint, quantify the gap, then choose the cheapest lever — stations, touch time, test capacity, or yield.

OP 55

How I Work

Clarity first.

If the floor is guessing, the process is already broken.

Standards over heroics.

A boring, reliable process beats a hero setup only one person can run.

Design for execution.

Plans are built around real setups, fixtures, tools, and inspection — not slideware.

Data over adjectives.

Escalations come with numbers. The MES exists so arguments end faster.

Feedback loops.

Machinists, brake operators, welders, and inspectors improve my process — every run.

People make the rate.

Staff to takt, set clear goals, protect the team from crisis-driven overtime.

OP 57

If It Doesn't Exist — I Build It

If it doesn't exist, or it doesn't work with your system, I build it.

I've written an MES from scratch, a machine health scoring model, and a metrics program with its own validation rules, a scheduling system, and physics engine- — not because I wanted a software project, but because the floor needed something and waiting wasn't going to work.

This pattern is one I keep coming back to: don't alarm on a single bad sample. Whether it's a spindle, a tank, or a telemetry channel, one noisy reading isn't a problem — a trend is. Persistence counts kill false alarms; slope tells you it's coming before the limit is crossed.

telemetry_monitor.py

# Limit checking with persistence and rate of change.

from collections import deque, defaultdict

class LimitMonitor:

    def __init__(self, limits, persistence=3, window=20):
        self.limits = limits        # ch: (lo_red, lo_yel, hi_yel, hi_red)
        self.persistence = persistence
        self.history = defaultdict(lambda: deque(maxlen=window))
        self.streak = defaultdict(int)

    def classify(self, ch, value):
        lo_r, lo_y, hi_y, hi_r = self.limits[ch]
        if value <= lo_r or value >= hi_r: return "RED"
        if value <= lo_y or value >= hi_y: return "YELLOW"
        return "GREEN"

    def slope(self, ch):
        h = self.history[ch]
        if len(h) < 2: return 0.0
        (t0, v0), (t1, v1) = h[0], h[-1]
        return (v1 - v0) / max(t1 - t0, 1e-6)

    def ingest(self, ch, t, value):
        # returns an event only when the streak trips
        self.history[ch].append((t, value))
        state = self.classify(ch, value)
        self.streak[ch] = self.streak[ch] + 1 if state != "GREEN" else 0

        if self.streak[ch] == self.persistence:
            return {"channel": ch, "state": state,
                    "value": value, "slope": self.slope(ch),
                    "t": t}
        return None

Same logic runs in my production systems — red/yellow limits, trend before threshold.

Building hardware that has to ship on schedule?

I'm in Southern California, working with teams that need development hardware turned into predictable production. Let's talk about your rate problem.

stevenjmcclain@hotmail.com ↓ DOWNLOAD RESUME (PDF)