Confirmed bar
JD and cohort agree
- End to end architecture and delivery
- Complex business logic and data systems
- Production APIs and cloud ownership
- High agency in a distributed team
Independent recruiting work sample Rebuilt 19 Jun 2026
Senior/Staff Backend Engineer at DualEntry
Talent market map,
with the uncertainty left in.
I used a live backend search to show how I calibrate on the team a company already hired, test adjacent talent pools, and turn the result into a sourcing and hiring system.
- usable calibration profiles
- 4
- external companies probed
- 7
- company pools verified
- 2
- research calls used
- 57 / 80
This work is independent and is not affiliated with or endorsed by DualEntry. Public professional names appear only as verification anchors. No candidate contact data or private CV details are included.
01 / Incumbent calibration
What the visible team suggests good looks like.
The calibration uses four usable in-role profiles. That is enough to form hypotheses, not enough to pretend the pattern is settled.
Low confidence
Two profiles are clearly backend leaning. One is a recent alumnus with thin project detail. One is an adjacent Senior Software Engineer with frontend and Node.js scope.
- 01
Architecture ownership
The strongest visible profiles have owned systems and tradeoffs, not only feature delivery.
- 02
Correctness-heavy systems
Financial integrations, payments, subscriptions, migrations, and data integrity recur more than one employer pedigree.
- 03
Scope over title
Prior leadership, platform scaling, and cross-system ownership are stronger signals than a literal Staff title.
- 04
Distributed delivery
The visible cohort spans European geographies and suggests written clarity and timezone discipline are part of the bar.
JD delta
The role is not simply asking for more of the same.
Likely hiring delta
More explicit in the JD
- Python backend depth
- PostgreSQL, schemas, ORMs, and migrations
- CI/CD, monitoring, and post release care
- Accounting logic and external integrations
Unwritten signals
More visible in the cohort
- Prior architecture or engineering leadership
- Applied AI, data, or event driven systems
- European distributed delivery
- Range across product, data, and infrastructure
02 / Talent Orbit
Seven companies tested. Two verified.
A company only advances when it appears through observed talent flow and an independent lookalike test, then produces a current person with at least two calibration signals.
Target DualEntry
VerifiedGuardian
VerifiedCockroach Labs
ParkedOmetria
ParkedConnect Earth
ParkedElsevier
ParkedmerXu
ParkedStar
Stop decision: 2 of 7 probed companies verified, a 0.29 relevance rate. The method stops below 0.30, so I did not spend the remaining research budget making the map look fuller.
Verified pools
Guardian News and Media
Flow + attributeSubscriptions, invoicing, price migrations, payments, and high traffic backend services.
Verification anchor: Pawel Krupinski, Senior Backend Engineer
Cockroach Labs
Leadership flow + attributeDistributed SQL, database migrations, change data capture, observability, and cloud infrastructure.
Verification anchor: Ryan Luu, Member of Technical Staff, Technical Lead
Parked after probing
- OmetriaStrong technical match. No equivalent current person verified after two probes.
- Connect EarthStrong finance-data match. Public engineering footprint was too thin.
- ElsevierRelevant semantic data work. People searches did not produce an equivalent anchor.
- merXuObserved incumbent flow. Public engineering evidence was too thin.
- StarObserved incumbent flow. Name collision and consultancy breadth weakened verification.
03 / Live market sizing
The evidence map and the reachable pool are different views.
Talent Orbit tests fit. LinkedIn audience estimates size the available search surface. I use both, but I do not turn discovery results into fake market counts.
Tier 1 Domain-led core
~2,400
Accounting, ERP, payments, ledger, reconciliation, and financial-data companies.
Tier 2 AI product signal
~8,300
Backend engineers with AI product signals, plus a ~840 named-company precision slice.
Tier 3 Regional scale
~36,000
Senior remote backend pool across the target regions, before domain calibration.
Directional LinkedIn audience estimates from the original search build. Rounded and not additive. The Tier 2 named-company precision slice was approximately 840 profiles.
Tier 1 sequence
Start narrow enough to learn.
Bucket 3 carries the volume. Buckets 1 and 2 should teach the first outreach and screen rubric.
Accounting / ERP
~660Lowest ramp on accounting workflows and ERP pain.
Source firstLedger / payments
~490Strong correctness instincts and a tighter pool.
Source secondFintech infrastructure
~2,300More scale, with a sharper motivation screen.
Use selectively04 / First sourcing pass
Nineteen profiles, graded against the work.
Company affiliation was not enough. The useful signals were contact with ledgers, reconciliation, e-invoicing, payroll, order to cash, migrations, or production AI on top of real backend depth.
- Strong
- 9
- Maybe
- 6
- AI companion
- 1
- Reject
- 3
Card 01
Direct ledger + AP/AR
- Signal
- Built a double-entry ledger as an audit-grade source of truth, plus AP/AR and invoice automation.
- Screen for
- Confirm Staff scope and availability.
Name and link held back
Card 02
E-invoicing + reconciliation
- Signal
- Owns tax compliance inside a global ERP, with prior bank and provider reconciliation.
- Screen for
- Confirm hands-on IC appetite versus management track.
Name and link held back
Card 03
Production AI companion
- Signal
- Shipped production LLM features with deep backend and platform experience.
- Screen for
- Confirm finance motivation and accounting-domain ramp.
Name and link held back
05 / Search operating system
The map has to change what happens on Monday.
Research is useful when it produces a tighter first week, a structured funnel, and feedback data that changes the next sourcing pass.
First week
- 01
Calibrate the backend archetype and Staff scope with the hiring manager.
- 02
Source the verified pools and Tier 1 accounting and ledger buckets first.
- 03
Review 30 to 40 profiles, then recut the map using observed rejection reasons.
- 04
Test two outreach angles: accounting correctness and database migration ownership.
- 05
Track source, response, screen pass, and reasons from the first profile.
Ashby-ready pipeline
I have not run Ashby in production. This design is based on equivalent workflows in Workday, Eightfold, Beamery, BambooHR, Airtable, Asana, and n8n.
- 01Sourced
- 02Recruiter screen
- 03Hiring manager
- 04Technical assessment
- 05Structured panel
- 06Founder / bar raiser
- 07Offer
- 08Closed with reason
Review from day one
- Source to reply
- Reply to screen
- Screen pass by pool
- Stage aging
- Reason-coded rejection
- Candidate response time
06 / What I would verify first
The questions that would recut the map fastest.
In work-sample mode, assumptions stay visible. These are the questions I would take into the first hiring-manager calibration.
- 01
Is this hire meant to add Python and PostgreSQL depth, or can equivalent distributed-systems depth clear the bar?
- 02
Which failure modes matter most: accounting correctness, migration safety, API reliability, integration edge cases, or throughput?
- 03
What observable scope separates Senior from Staff in this team?
- 04
How many hours of daily EST overlap are actually required?
- 05
Does direct ERP experience change interview conversion, or are payments, subscriptions, and migrations accepted substitutes?
- 06
Which current or recent engineer best represents a strong yes for this exact role?
The broader work
This is how I keep a hard search from becoming a pile of profiles.
Calibrate on evidence. Separate market size from market fit. Make uncertainty visible. Then build the workflow that lets the team learn faster than the search changes.