Levels and step-up matrix · FDE Level Standard

L1Entry · Execution · TasksJunior Delivery Engineer / Junior Business Analyst

Delivers one clearly bounded piece of work correctly, inside an existing framework and template.

Step-up at this level	Baseline: from “can use the tools” to “can hand over one clean piece of work”.
Typical scope	Scenarios: A single task inside one scenario (one module, one step, one interface); Systems: 1–2 modules within one system; Stakeholders: Front-line users and a mentor; Horizon: Days to weeks; Budget: No budget ownership
Capability marks	Business literacy Guided / Scenario engineering Guided / AI engineering and debugging Guided / Delivery assurance Guided / Delivery engineering Guided
Observable evidence	Runs the existing template unaided and can explain what each step does Keeps a record of failures and fixes, not just a screenshot of success Can say where the work is most likely to break when handing over
Deliverables	A working module or configuration · Start-up notes · Failure sample log
Metric language	Task-level effort and rework count (minutes / hours)
Promotion criteria	Three consecutive tasks delivered without rework Judges correctly when to pause and ask for help At least one document reused directly by someone else
Fast-track	No fast-track at this level. The point is the fundamentals.
Common misconception	Mistaking tool fluency for delivery capability.
Failure mode	Work is finished but not handed over cleanly: no documentation, no failure samples, the next person starts from zero.

L2Intermediate · Independent · ScenariosDelivery Engineer / Scenario Delivery Consultant

Owns one scenario end to end: scope, baseline, acceptance, and the system still running afterwards.

Step-up at this level	From delivering a task to owning an outcome — you start signing your name to a process.
Typical scope	Scenarios: One scenario, full cycle (from requirement to live operation); Systems: Data and interfaces across 1–3 systems; Stakeholders: Business owner, front-line staff, IT / data team; Horizon: Weeks to months; Budget: Single-scenario scale, subject to approval
Capability marks	Business literacy Independent / Scenario engineering Independent / AI engineering and debugging Independent / Delivery assurance Independent / Delivery engineering Independent
Observable evidence	A scope sheet: boundaries, baseline numbers, acceptance actions A failure sample set (20+ cases) that shows where the system breaks A record of one real process running in production, with named users
Deliverables	A running scenario · Acceptance sheet · Failure sample set · First version of the runbook
Metric language	Scenario metrics (accuracy, cycle time, handle time, error rate)
Promotion criteria	Two scenarios delivered independently, one of them running for 4+ weeks Defines the baseline and writes acceptance as a checkable condition One deliverable reused by a second scenario
Fast-track	Equivalent-scale evidence (including acceptance and handover) can skip the task-level review, but never the evidence review.
Common misconception	Treating go-live as delivery complete. Go-live is the start.
Failure mode	The scenario works but nobody uses it: acceptance measured features, not outcomes.

L3Advanced · Coordination · Project groupsSenior Delivery Engineer / Senior Scenario Consultant

Coordinates a portfolio of scenarios and resources, turning one hard win into a repeatable playbook.

Step-up at this level	From doing one well to deciding which ones to do — the first time you hold the trade-off.
Typical scope	Scenarios: A portfolio of 3–10 scenarios, with prioritisation and trade-offs; Systems: Cross-system, cross-department orchestration; Stakeholders: Department heads and above, multiple vendors; Horizon: Quarterly; Budget: Portfolio scale, participates in budget planning
Capability marks	Business literacy Adaptive / Scenario engineering Adaptive / AI engineering and debugging Adaptive / Delivery assurance Adaptive / Delivery engineering Adaptive
Observable evidence	A portfolio plan: why these three first, why those five were dropped A repeatable review and operations mechanism (who checks what, when) At least one cross-department conflict resolved with a documented trade-off
Deliverables	Scenario portfolio roadmap · Review and operations mechanism · Architecture decision records · Two or more reusable assets
Metric language	Portfolio value (total cost, total hours, qualification rate, reuse rate)
Promotion criteria	Coordinates 3+ scenarios, all reaching acceptance Built a mechanism or asset used by others at least twice Makes an explainable trade-off under resource pressure instead of doing everything
Fast-track	L4-grade assetisation on a single complex system (reused across 3+ projects) can fast-track, subject to unanimous review.
Common misconception	Mistaking meeting volume for coordination ability.
Failure mode	Everything in progress, nothing finished: no priorities means no priority.

L4Lead · System · Organisational systemsLead Delivery Engineer / Head of Scenario Delivery

Designs the delivery system of the organisation: quality gates, assetisation, capability building — outcomes that do not depend on one person.

Step-up at this level	From delivering yourself to making the organisation able to deliver — output shifts from projects to mechanisms.
Typical scope	Scenarios: System design across project groups (10+ scenarios, multiple business lines); Systems: Platform and toolchain, including reuse architecture and quality gates; Stakeholders: Executive sponsors, cross-company partners; Horizon: Six months to a year; Budget: System scale, leads budget preparation
Capability marks	Business literacy Systemic / Scenario engineering Systemic / AI engineering and debugging Systemic / Delivery assurance Systemic / Delivery engineering Systemic
Observable evidence	A delivery SOP and quality gate that is actually executed, with records An asset library with reuse data (reuse counts, hour savings) At least one person brought to independent scenario ownership, with a handover record
Deliverables	Delivery system and gates · Asset library and reuse metrics · Capability development path · Cross-project retrospective mechanism
Metric language	Business language (capital tied up, unit delivery cost, capitalised assets, margin impact)
Promotion criteria	System in force for two quarters with a measurable drop in incidents and rework Asset reuse cuts similar-project hours by 30% or more Developed at least one L2+ person with independent delivery evidence
Fast-track	An externally adopted method or standard can enter L5 review directly, with adoption evidence.
Common misconception	Mistaking headcount for system capability — the test is whether results survive turnover.
Failure mode	Beautifully written policy nobody follows: gates are decorative and problems are still absorbed by heroes.

L5Expert · Method · Industry methodsDelivery Expert / Scenario Expert

A method adopted beyond one company: standards, ecosystem and cross-enterprise transfer.

Step-up at this level	From working inside one organisation to outsiders adopting it — judgement moves outside.
Typical scope	Scenarios: Cross-industry method output (standards, curricula, open-source assets); Systems: Industry-level methodology and platform ecosystem; Stakeholders: Industry bodies, ecosystem partners, regulators and standards committees; Horizon: One year and beyond; Budget: Ecosystem scale, involves external investment
Capability marks	Business literacy Industry / Scenario engineering Industry / AI engineering and debugging Industry / Delivery assurance Industry / Delivery engineering Industry
Observable evidence	A method adopted by an external body or peer (citation, adoption, joint publication) Two or more cross-enterprise transfers (not within the same group) A publicly verifiable standard, curriculum, open-source asset or signed report
Deliverables	Industry standard or methodology · Ecosystem collaboration mechanism · Publicly citable research
Metric language	Industry impact (adoption count, ecosystem partners, shift in industry practice)
Promotion criteria	Method adopted by 2+ enterprises outside the group with verifiable results A substantive role in an industry body or standards effort Developed at least one L3+ person
Fast-track	Judged only by external adoption. Self-certification is not accepted; no external evidence means no level.
Common misconception	Mistaking speaking volume for influence — influence is others copying the method.
Failure mode	Method stays in talks and articles: no external enterprise has actually run it.

Step-up matrix: five domains × five levels

Read across for a domain’s progression, down for a level’s full picture. If a column looks like the previous one, that level has no real step-up.

Scroll horizontally on small screens.

Domain	L1 Entry	L2 Intermediate	L3 Advanced	L4 Lead	L5 Expert
Business literacy	Guided Records the process and roles using a given outline without errors	Independent Runs one process walkthrough alone and states inputs, outputs and owners	Adaptive Spots conflicting metrics and pseudo-requirements, and finds the real bottleneck	Systemic Judges where AI applies on a business line and turns the judgement into a reusable checklist	Industry Publishes an industry problem framework that peers adopt
Scenario engineering	Guided Completes scope and acceptance from a template without widening it	Independent Confirms scope, measures the baseline, writes checkable acceptance actions	Adaptive Prioritises a portfolio of scenarios and explains what was dropped	Systemic Defines organisational admission and acceptance standards and enforces them in review	Industry Outputs scenario classification used by outsiders
AI engineering and debugging	Guided Reproduces correct and failing output from a given script and records the evidence	Independent Builds a 20+ failure sample set and separates retrieval faults from generation faults	Adaptive Resolves model / data / tool interaction faults and designs the fallback	Systemic Builds evaluation and regression mechanisms and turns quality into a gate	Industry Publishes an evaluation method or quality metric others adopt
Delivery assurance	Guided Reports progress and completes handover from a checklist without dropping known risks	Independent Connects data and permissions, explains minimisation and masking, drives pilot and takeover	Adaptive Coordinates dependencies and change, controls scope creep and cost overrun	Systemic Establishes data governance, incident response and delivery quality mechanisms	Industry Contributes to industry delivery or compliance practice
Delivery engineering	Guided Collects reusable snippets with instructions others can follow	Independent Produces at least one asset (connector, template, evaluation set) used in a second place	Adaptive Designs the reuse interface, accumulates 2+ assets and measures reuse	Systemic Builds an asset library with reuse metrics and a visible drop in project hours	Industry An asset is adopted by an external company or reused in the open

Six evidence packs

What must be produced at each level. Any empty pack means the evidence is incomplete.

Scroll horizontally on small screens.

Evidence pack	L1	L2	L3	L4	L5
Business diagnosis	Records the user process and pain points without missing key roles	Describes the problem in the customer’s metrics and gives a quantified baseline	Separates real pain points from pseudo-requirements across departments	Judges where AI applies on a business line and states the trade-off	Defines an industry problem framework others adopt
Scenario and acceptance	Fills scope and acceptance from the template without widening it	Confirms scope, baseline and acceptance actions independently	Ranks a portfolio of scenarios and allocates resources	Defines organisational admission and acceptance standards	Publishes a scenario classification peers cite
Data and permissions	Connects data under guidance and does not touch permission settings	Connects data and permissions and explains minimisation and masking	Designs cross-system data flows and permission tiers	Builds data governance and audit mechanisms	Contributes to industry data and compliance practice
AI debugging and evaluation	Reproduces both correct and failing output	Builds a 20+ failure set and separates retrieval from generation faults	Designs the evaluation system and handles interaction faults	Builds regression mechanisms and quality gates	Publishes an evaluation method the industry adopts
Delivery operations	Hands over from a checklist and reports progress	Drives pilot, acceptance and user takeover	Coordinates dependencies and change, controls scope creep	Builds delivery quality and incident response mechanisms	Advances industry delivery practice
Assetisation	Collects reusable snippets and instructions	Produces one reusable asset used in a second place	Designs the reuse interface, accumulates 2+ assets and measures reuse	Builds an asset library with reuse metrics and quantified hour savings	An asset is adopted externally or reused in the open

Big-tech level ranges

Compiled from public information · not an official mapping · these systems have been reformed repeatedly. Ranges only, no point values.

Level	Alibaba	ByteDance	Tencent	Baidu	Confidence
L1 Entry	P4–P5	1-1 ~ 2-1	T1.3–T2.1	T4–T5	Medium
L2 Intermediate	P5–P6	2-1 ~ 2-2	T2.1–T2.3	T5–T6	High
L3 Advanced	P6–P7	2-2 ~ 3-1	T2.3–T3.1	T6–T7	Medium
L4 Lead	P7–P8	3-1 ~ 3-2	T3.1–T3.3	T7–T8	Medium
L5 Expert	P8–P9	3-2 ~ 4-1	T3.3–T4	T8–T9	Low

Market salary bands

Annual salary in CNY thousands, gross, including bonus, excluding equity. Basis: job postings × 12–15 months, cross-checked against salary reports. Not percentiles, not what any employer pays.

Level	Suzhou	Hangzhou	Shanghai	Beijing / Shenzhen	Basis
L1 Entry	12–20	14–23	16–26	18–30	Entry to 2 years; derived from job postings at CNY 12k–20k per month
L2 Intermediate	20–33	23–38	26–42	30–48	3–5 years; calibrated to the lower bound of the CNY 47k AI integration average
L3 Advanced	32–52	36–59	42–68	48–78	5–8 years; includes scenario-owner premium (AI skill premium +62%)
L4 Lead	50–85	56–95	65–110	75–130	8+ years; lower half of the senior architect band
L5 Expert	80–130	90–145	100–170	120–200	Very small sample; anchored to the lower bound of the CNY 1.5–2.0m senior band

Hangzhou has no separately published band; the figure is an estimate interpolated between Suzhou and Shanghai. All cities use the same job-posting and salary-report basis.

The full self-assessment is currently available in Chinese only. An English question bank is in preparation. Open the Chinese self-check