Marwan
Mechanical engineering
Understanding how parts work together. Paying attention to what happens when they don’t.Catch specification gaps before the purchase order.
Two documents. One traceable review.
Reports saved in this browser. Original PDFs are not saved; retain your documents separately.
Have an engineering task that takes longer than it should? Feedback on the beta? We’d love to hear from you.
mymarwaaannnnnny@myeng.onlineBorn from friendship. Built around the details.

Different disciplines.
A shared way of thinking.
We’re Marwan and Wadiah, the founders of MYENG. Our friendship began around seven years ago. We studied at different universities, but our subjects often brought us to the same table: Marwan in mechanical engineering, Wadiah in physics and renewable energy.
We worked through shared subjects, took engineering courses together, and kept noticing the same thing. Too much careful engineering work depended on repetitive checks, scattered information, and details that were easy to miss.
So we started building small tools for ourselves. Tools that made a task clearer. A calculation easier to check. A discrepancy harder to overlook. Precision mattered to us, and we wanted the tools we used to reflect that.
After graduation, a personal habit became a shared ambition: turn those tools into something other engineers could use, and build a sustainable business around work we care about.
That’s the idea behind MYENG. Thoughtful engineering tools that make the details easier to see, without taking the engineer out of the decision. Our first public beta focuses on comparing pump requirements with supplier datasheets—one specific problem, handled with care.
Mechanical engineering
Understanding how parts work together. Paying attention to what happens when they don’t.Physics & renewable energy
Looking closely at the principles behind a system. Finding clearer ways to work with them.A focused tool for the moment between receiving a datasheet and approving an order.
Try the review workspace ↗A project calls for a material, duty point, or pressure rating. A supplier proposes something different. MYENG places the two statements side by side so the discrepancy has a source, an owner, and a next step.
Local mode compares explicitly labeled text fields with unit conversions. Claude mode extracts the same eight fields from text or PDFs. Code compares numeric values; engineers verify the source evidence and resolve clarifications.
A match only covers the extracted field. It does not approve a pump, prove its duty-point performance, assess a full curve, or certify standards compliance. Missing fields, ambiguous statements, and unsupported units need review. PDF evidence is AI-reported until you check it against the document.
Local comparisons run in your browser. Claude mode sends both documents through MYENG’s server to Anthropic using your own session-only API key. API charges apply. Saved history stays in browser storage; original PDFs and keys are never saved there. Upload only documents you are authorized to share.
Materials: recognized spelling variants and generic family requirements only. CF8M / 1.4408 and cast/wrought substitutions require approved equivalence. Connections: no automatic DN-to-NPS or PN-to-ASME interchangeability. Pressure: the same rating definition, gauge/absolute basis, and conditions must be evidenced. NPSHr is a maximum at the same duty; this does not verify the available NPSH margin.
Reference context: Outokumpu stainless grades · KSB NPSH · Swagelok flange ratings. These references do not certify a supplier substitution.
The beta includes eight fields: flow, head, casing material, pressure rating, maximum temperature, discharge connection, motor power, and maximum NPSH required. It includes source evidence, versioned field rules, recorded engineer corrections, local audit events, clarification drafts, CSV export, and printable reports. Missing evidence is “Can’t verify,” never an all-clear. No customer adoption, certification, or automatic engineering approval is claimed. Accuracy on real anonymized document pairs has not yet been measured. 49 synthetic regression cases cover supported rules and uncertainty paths; they do not establish real-world recall or precision.