A four-phase delivery program — Phase 0 discovery, Pilot, Launch, and Expansion — engineered end-to-end by Astera Nova: architecture, integrations, product, compliance, and operations, built for an online-only bank from day one.
Headline scope and investment per phase. Full objective, prerequisites, assumptions, deliverables, and team breakdown for each phase are detailed further down the page.
Every bar below is a real work package with its own start and end month — nothing runs full-width. 19 of the 50 bars form the critical chain that actually sets the 16-month duration; the other 31 run alongside with real slack. The 21 arrows shown are only genuine hard blockers (X literally can't start until Y is done) — 12 documented relationships, some fanning out to more than one successor; dashed gates pin the 7 sync points to the exact month they gate.
Critical path (highlighted): SA architecture (P0-SA1/SA2) → Pilot connector & product build (PL-INT1/PROD1) → Pilot UAT (PL-QA2, Gate 4) → Launch hardening & catalog (LA-INT1/PROD1) → Launch E2E (LA-QA1/QA2) → pentest/compliance/DR converge (LA-SEC1, LA-COMP1, LA-DEVOPS2, Gates 5&6) → Cutover (M13) → Expansion build & test (EX-SA1/PROD1/QA1). 19 bars, 8 phase-to-phase serial links — this is what the 16 months actually depends on; the other 31 bars run in parallel with slack.
Legacy-Core API confirmation. SA resolves REST-vs-batch reality for GL/wallet APIs and secures sandbox access.Gate 1 · M3 → blocks Pilot connector (PL-INT1) at M4
Bureau/score-provider selection. Vendor (CIC/FinScore/other) chosen, sandbox access confirmed for Pilot integration.Gate 2 · M2 → blocks Pilot bureau connector & scorecard (PL-INT1, PL-CRED1) at M5
Bank KYC/AML ownership confirmation. Bank confirms it owns KYC/AML sign-off and the API shape it exposes.Gate 3 · M3 → blocks Pilot compliance controls (PL-COMP1) at M5
Pilot UAT sign-off. Bank-coordinated UAT complete, retrospective run, formal go/no-go recorded.Gate 4 · M8 → blocks Launch hardening/build (LA-INT1, LA-PROD1, LA-DEVOPS1) at M9
Compliance sign-off + pentest remediation. Production sign-off documented; pentest findings closed and verified.Gate 5 · M13 → blocks Production Cutover (LA-CUTOVER) at M13
DR/production-readiness verification. DR/backup infrastructure tested; failover/DR architecture verified against RTO/RPO.Gate 6 · M12 → blocks Production Cutover (LA-CUTOVER) at M13
Historical-data/ML scope decision. ~1,000-loan portfolio confirmed insufficient for production ML.Gate 7 · M3 → locks Launch scoring rules-based & Expansion ML exploratory-only
Establish the architectural, regulatory, and delivery foundation before any product code is built. Phase 0 answers the questions that determine Pilot's design: what the bank's legacy core/GL/wallet can actually expose via API, what BSP/CIC/AMLA/Data-Privacy rules constrain product and data design, and what the reference architecture, tech stack, and environments look like for all 9 components. No build — design, audit, and planning only.
| Role | What they do in Phase 0 | Hours | Cost |
|---|---|---|---|
| SA | Legacy-core API assessment (REST vs. batch) — program risk #1, gates Pilot's connector build. Scoring architecture that has to work rules-first, no ML at launch. RBAC/maker-checker model for the confirmed 10-role + 4-API-actor surface. | 724 | $94,120 |
| BA | Bureau/score-provider vendor audit (CIC vs. FinScore vs. other) — kicks off sandbox access, program risk #2. Compliance traceability matrix that draws the exact line between what we own and what the bank exposes via API. | 500 | $37,500 |
| COMP | AMLA/KYC analysis scoped to the bank's API contract, not an AML build — the "bank owns compliance operations" boundary the whole estimate depends on. AI/ML-in-scoring read stays light since launch scoring is rules-based. | 280 | $22,400 |
| DevOps | Discovery environment, IaC and CI skeleton — routine scaffolding, low risk this phase; sets the baseline Pilot's real environments build on. | 112 | $10,640 |
| DE | Legacy-core/GL/wallet inventory with hands-on API probing, not just document review — direct input to SA's #1-risk API assessment. | 96 | $8,640 |
| PM | Program charter and Phase 0 roadmap — sets the cadence for a 16-month, four-phase program with a lean team. | 88 | $7,920 |
| SEC | Threat modeling raises the remote-identity-proofing / account-takeover bar from day one — online-only bank, no branch fallback to catch a bad actor. | 64 | $7,040 |
| QA | Test-strategy and automation-tooling draft, scoped to a thin UI surface — the decision Pilot's QA either reuses or has to redo. | 40 | $2,600 |
| MLE | Historical-data/scoring feasibility on ~1,000 loans — the finding that locks Launch scoring rules-based rather than ML. | 32 | $3,200 |
| Phase 0 Subtotal | 1,952 h | $195,100 |
Build one thin, fully end-to-end slice — one term-deposit product and one unsecured-loan product — through every one of the 9 components: origination, scoring (rules-based), servicing, collections, operations/RBAC, compliance controls, and the core/GL/bureau/KYC/wallet integrations. Prove the architecture against a real bank sandbox before committing to full-scale Launch build.
| Role | What they do in Pilot | Hours | Cost |
|---|---|---|---|
| BE | Core-banking/GL/bureau connector build against an API shape only confirmed at the Gate 1/2 handoff — the single biggest execution risk in the phase. RBAC + maker-checker across the full 10-role model, not a simplified pilot stand-in. Reconciliation-aware GL posting, not just "post a transaction." | 1,568 | $141,120 |
| DevOps | Full dev/test/staging + CI/CD stood up before bank sandbox access is guaranteed — a sequencing risk if that access slips. | 352 | $33,440 |
| FE | Maker-checker approval queue UI — the one screen tightly coupled to backend RBAC semantics, not a generic CRUD form. | 340 | $28,900 |
| QA | Builds the one test-automation framework the program reuses through Launch and Expansion, not a pilot-only throwaway — targeted at the few real interfaces and the core connector/contract flows, not broad manual UI passes. | 120 | $7,800 |
| PM | Coordinates the bank stakeholders and runs the pilot retrospective that directly feeds the Launch plan. | 200 | $18,000 |
| DE | Test-data prep for interest day-count logic has to mirror the legacy core's actual accrual convention — unconfirmed until Gate 1 closes. | 180 | $16,200 |
| SA | API contracts between components — the artifact that turns Phase 0's architecture assumptions into commitments Pilot code is actually built against. | 176 | $22,880 |
| BA | Documents scoring business rules — the record of exactly which decisions are rules-based today, in case the ML add-on gets greenlit later. | 168 | $12,600 |
| MLE | Rules-based champion scorecard, deliberately not an ML model. The real work is building drift/PSI monitoring for a system with no live production data yet. | 144 | $14,400 |
| SEC | First threat model against live connectors and real secrets — still shaped by the no-branch, remote-only posture. | 136 | $14,960 |
| COMP | Basic KYC/AML controls scoped strictly as bank-API consumption — the calibration boundary made concrete for the first time, not a parallel control build. | 48 | $3,840 |
| Pilot Subtotal | 3,432 h | $314,140 |
Take the validated pilot to full production coverage across all 9 components: harden every connector and calculation path, extend the product catalog and rules to the full initial launch scope, complete regulatory reporting and compliance sign-off, and prove production readiness through full end-to-end testing, load testing, and a pre-launch pentest.
| Role | What they do in Launch | Hours | Cost |
|---|---|---|---|
| BE | Hardens connectors for retry/idempotency/reconciliation — the gap between a pilot integration and a production one. Extends eligibility/affordability/fraud rules to the full catalog without re-architecting Pilot's single-product logic. | 888 | $79,920 |
| QA | Automated E2E across the 9 components' contracts/critical flows, not exhaustive manual UI regression — the thin interface surface is exactly why this stays lean. Load/performance testing holds (non-negotiable for a live-money system). Security testing runs jointly with SEC, not sequentially after. | 150 | $9,750 |
| DevOps | DR/backup build & test is a hard release gate (Gate 6) — parallel to the pentest and compliance sign-off, all three have to close together before cutover. | 416 | $39,520 |
| BA | Full documentation and bank-team training — the handover artifact that determines how much the bank can self-serve after go-live. | 328 | $24,600 |
| SEC | Pre-launch pentest + remediation is a hard release gate (Gate 5) — findings block cutover, this isn't a checklist review. | 304 | $33,440 |
| PM | Runs cross-component dependency management across all 9 components converging on one cutover date. | 288 | $25,920 |
| DE | Reconciliation reports against connectors still being hardened in parallel — the pipeline can't fully stabilize until connector hardening lands. | 244 | $21,960 |
| FE | Extends the admin UI to the full parameter set — mostly scale, not new risk. | 224 | $19,040 |
| SA | Failover/DR strategy design — the architecture half of the Gate 6 DR-readiness gate. | 168 | $21,840 |
| COMP | Production compliance sign-off converges with pentest remediation and DR verification at the same two-week cutover window — three independent tracks landing together. | 144 | $11,520 |
| MLE | Challenger-model stability monitoring — first time the scorecard runs unattended at production volume. | 120 | $12,000 |
| Launch Subtotal | 3,274 h | $299,510 |
Deliver the first post-launch increment: 2–3 additional deposit variants, 1–2 additional unsecured-loan products, expanded eligibility/affordability data sources, promo mechanics, and the first collections risk-scoring pilot — while validating the platform scales cleanly beyond the initial launch catalog. Explicitly scoped as the first increment, not an open-ended roadmap.
| Role | What they do in Expansion | Hours | Cost |
|---|---|---|---|
| BE | Optional second bureau connector and SSO/AD integration — both explicitly unconfirmed-scope items, not committed builds; scope depends on bank decisions still open. | 612 | $55,080 |
| BA | New eligibility rules and promo-campaign design for the confirmed 2–3 deposit / 1–2 loan first wave. | 168 | $12,600 |
| DE | Feature-engineering pipeline groundwork for the eventual ML add-on — not the model itself, just the plumbing. | 144 | $12,960 |
| MLE | Champion-vs-ML-challenger A/B groundwork — exploratory only. This is the seed of the separately-priced ML add-on, not a production model. | 144 | $14,400 |
| FE | Promo-configuration UI — routine extension of the existing admin UI pattern. | 96 | $8,160 |
| QA | Regression pass reuses Launch's full suite rather than rebuilding — the same reuse pattern behind the QA-hours recalculation above. | 40 | $2,600 |
| DevOps | Scales infrastructure for the first wave — routine, no new architecture. | 80 | $7,600 |
| SA | Architecture for the first wave of additional products, reviewed against the production baseline set at Launch. | 64 | $8,320 |
| COMP | First periodic control audit — the compliance cadence shifts from one-off sign-off to ongoing review. | 60 | $4,800 |
| PM | Manages expansion-wave dependencies — smallest PM line in the program, reflecting one confirmed wave, not an open-ended roadmap. | 56 | $5,040 |
| SEC | Security review scoped to the new components only — not a full re-pentest. | 32 | $3,520 |
| Expansion Subtotal | 1,496 h | $135,080 |
Optional ML scoring add-on (not in the above hours): a dedicated, production-grade ML credit-scoring uplift — model build, MLOps, governance, bias/fair-lending testing — is available as a separate opt-in track, top-of-market ballparked at 500–700h, additive to the exploratory groundwork above.
Every role's program-wide mandate, and its total hours/cost summed across all four phases — the same 10,138 h / $942,790 reconciled from a different angle: by role instead of by phase.
Blended rate $92.99/h. Rates: SA $130 · BA $75 · BE $90 · FE $85 · DE $90 · MLE $100 · QA $65 · DevOps $95 · PM $90 · COMP $80 · SEC $110.
| Role | Program-Wide Mandate | Total Hours | Total Cost |
|---|---|---|---|
| BEBackend Engineer | The largest role on the program. Owns every server-side build: product/origination/servicing logic, all core-banking/GL/bureau/KYC/wallet connectors, the scoring decision engine, RBAC/maker-checker, and regulatory report generation — from Pilot's single-product slice through Launch's full catalog and Expansion's new products. | 3,068 h | $276,120 |
| BABusiness Analyst | The process and requirements backbone: audits current processes, designs traceability matrices and product-policy models, writes documentation/runbooks/training, configures product parameters — active every phase from discovery through expansion design. | 1,164 h | $87,300 |
| SASolution Architect | Owns the reference and integration architecture end to end — heaviest in Phase 0 (foundational design) and Launch (production hardening/DR), lighter touch in Pilot (API contracts) and Expansion (scaling for new products). | 1,132 h | $147,160 |
| QAQuality Assurance | Owns all testing beyond unit tests, scoped tightly for a thin-interface product: one automation framework built once and reused every phase, critical-flow/contract checks over broad manual UI regression, load/performance and security testing held firm as hard release gates. | 350 h | $22,750 |
| DevOpsDevOps Engineer | Builds and hardens platform infrastructure: environments, IaC, CI/CD, observability, DR — scaling from a discovery sandbox in Phase 0 to production-grade blue-green infrastructure at Launch. | 960 h | $91,200 |
| DEData Engineer | Builds the data pipeline and analytics layer: integration/data-source audits in Phase 0, transactional/analytics pipelines in Pilot, portfolio analytics and reconciliation at Launch, ML-adjacent feature-pipeline groundwork in Expansion. | 664 h | $59,760 |
| FEFrontend Engineer | Builds every admin and operator-facing UI: product/role administration, application processing, collections dashboards, business dashboards — extended in Launch and Expansion for the full product set. | 660 h | $56,100 |
| PMProgram Manager | Runs program governance end to end: mobilization/charter in Phase 0, pilot management in Pilot, full launch-program and cross-component dependency management at Launch, expansion-wave planning. | 632 h | $56,880 |
| SECSecurity Engineer | Owns the security posture: top-level threat modeling/SDLC standards in Phase 0, pilot secure-coding review, pre-launch pentest and remediation plus infra hardening at Launch, expansion component security review. | 536 h | $58,960 |
| COMPCompliance | Owns regulatory analysis and sign-off support: BSP/AMLA/CIC/Data-Privacy analysis in Phase 0, basic product controls in Pilot, production compliance sign-off and disclosure validation at Launch, periodic control audits in Expansion. | 532 h | $42,560 |
| MLEML Engineer | Scoped light and rules-first per program decision: data/feasibility assessment in Phase 0, rules-based champion scorecard in Pilot, scorecard extension/monitoring at Launch, exploratory ML/A-B groundwork in Expansion — a production ML uplift remains a separate opt-in track. | 440 h | $44,000 |
| Program Total | 10,138 h | $942,790 |