From chemistry to code: what a science background brings to systems analysis
How training as a scientist shapes the way I decompose, document and de-risk technical delivery.
Read article07 — Insights
Notes on shipping software.
Field notes on delivery, requirements and rescuing stalled products — written for the people who have to make systems work.
SSADM Is Not Dead — But the Full Lifecycle Probably Is
The logical data model in the shared drive still names the entities your team argues about — even if nobody has said “SSADM” in a decade.
Read on MediumThe Apology That Fixed Nothing
A templated sorry after a small miss repairs your discomfort, not your team's trust — and the wrong script can cost you more than silence.
Read articleManaging Up When You Don't Perform Confidence
Skeptical sponsors don't need your assurance tone — they need a ledger they can audit.
Read articleBalanced Random Forest on Network Flows — When SMOTE Is Worth the Disk
Manual oversampling is feasible on training data, but embedded balancing is usually the first experiment that survives a million benign rows.
Read on MediumHigh Output That Isn't Sustainable — the Analyst Edition
Your dashboard says you're winning. Your capacity might disagree.
Read articleWhat an AI Summary Removes From a Requirements Conversation
The handover looked complete. The transcript still held the condition.
Read articleWhen Compliance Training Meets How People Actually Work
The completion badge on your LMS and the shared spreadsheet on the file server are both telling the truth — about different things.
Read articlePurple Team Is Not a Third Squad — It's How Red and Blue Prove Detection Works
Purple isn't a separate team — it's the practice where red and blue run the same attack together until the detections actually fire.
Read on MediumSelf-Efficacy When the Spec Was Wrong Before You Arrived
Inherited requirements drain agency before they drain skill — and the fix is smaller than a rewrite.
Read articleInsider Threat and AI — The Leak Path Left Your DLP Behind
Your DLP still watches file size. Your analysts paste customer data into chat boxes. Those two facts don't reconcile.
Read on MediumRescuing a stalled software project: a 6-month playbook
The repeatable steps I use to take a product that's been stuck for a year and get a shippable MVP out the door.
Read articleRequirements that survive contact with reality
Why most requirements documents fail in week three — and how to write ones that hold up when the build starts.
Read articleLegacy code is a risk register, not a birth certificate
Install date is the wrong sorting key. Score the signals that predict where change actually hurts — change friction, knowledge concentration and social debt.
Read articleSmall business AI is a growth lever, not a layoff plan
Census and Main Street payroll data show AI replaces tasks, not jobs by default — and how small firms reinvest the recovered capacity.
Read articleHiring for Brownfield Temperament — What to Listen For
Your loop screens for trivia and rapport; the job needs someone who can live inside ambiguity without reaching for a rewrite.
Read articleAI Deepfake Phishing Didn’t Get Smarter — It Got Cheaper to Believe
Believability is the exploit now — and most security programs still train people to spot typos in email.
Read on Medium// Articles in progress — subscribe or connect on LinkedIn to be notified when they go live.