ICT-Platform Dev Specialist FA study plan
ICT-Platform Dev Specialist FA · Give each week a clear focus.
8 weeks · About 7 hours per week
Platform Architecture and System Design
Map every layer of a reference platform (edge/CDN, gateway, service tier, async workers, data stores, identity provider) into a single annotated diagram you draw from memory at the end of the week
Compare monolithic, layered, microservice, and event-driven styles by listing two concrete trade-offs each for latency, operability, and failure isolation, then justify a style choice for a given workload
Explain synchronous request/response versus asynchronous queue or stream processing with a worked latency and backpressure example for each
Write the non-functional requirements (availability target, p99 latency budget, RPO/RTO) for a sample platform and trace each one to an architectural decision
Your checkpoint
A one-page architecture decision record for a fictional internal platform, including component diagram, chosen style, and three rejected alternatives with rationale
Platform Architecture and System Design & Application Development and Programming
Decompose a platform into bounded services, defining each service's responsibility, owned data, and public contract in a table
Design the data flow for one cross-service workflow, specifying where transactions end and where eventual consistency with compensation begins
Build a small working service (any mainstream language) exposing two REST endpoints backed by an in-memory store, with input validation and structured logging
Refactor that service to separate transport, business logic, and persistence layers, and confirm behavior is unchanged via tests
Your checkpoint
A service decomposition table plus a running two-endpoint service in a public repository with at least four passing unit tests
Application Development and Programming
Implement the same small feature in a stateless and a stateful style, then document which one survives horizontal scaling and why
Add centralized error handling and idempotency keys to the week-2 service so a retried write does not duplicate records
Write integration tests that exercise the service through its HTTP surface against a containerized dependency
Profile one endpoint and identify the single largest contributor to its response time
Your checkpoint
A pull request adding idempotent writes, centralized error handling, and three integration tests, with a short profiling note naming the measured bottleneck
Data Modeling, Storage, and Database Management
Model a realistic domain in both normalized (third normal form) and deliberately denormalized form, and state the read/write pattern that justifies each
Write and explain the execution plan for five queries: one using a composite index, one causing a full scan, one with a join across three tables, one with pagination, and one with an aggregate
Implement optimistic concurrency control with a version column and demonstrate a conflicting update being rejected
Choose between relational, document, and key-value storage for three given access patterns and defend each choice in two sentences
Your checkpoint
A schema file with migrations, indexes, and a written query-performance report covering the five explained queries with before/after timings
API Design, Integration, and Interoperability
Design a versioned REST API for one resource, specifying status codes, error envelope, pagination, filtering, and idempotency behavior
Write an OpenAPI description for that API and generate a client or mock server from it to prove the contract is complete
Implement one backward-compatible change and one breaking change, and describe the deprecation and migration path for the breaking one
Add authentication to the API and demonstrate a request failing with 401 and another succeeding with a valid token
Your checkpoint
A published OpenAPI specification plus a short compatibility note listing the breaking change, its migration steps, and the sunset window
Cloud, Infrastructure, and Deployment Operations & Security, Identity, and Access Control
Containerize the service from earlier weeks with a multi-stage build and run it alongside its database using local orchestration
Define infrastructure as code for the service, its database, and network boundaries, then destroy and recreate the environment from that definition alone
Implement role-based access control with at least three roles and verify each role against one allowed and one denied action
Store all secrets outside source control and prove the running service reads them from the environment or secret manager
Your checkpoint
A reproducible infrastructure definition that provisions and tears down the full stack, plus an access-control test matrix showing pass/fail per role
Performance, Reliability, and Monitoring
Instrument the service with structured logs, request tracing, and at least three metrics covering latency, error rate, and saturation
Run a load test to find the throughput ceiling and record where latency degrades past the stated budget
Define service-level indicators and objectives for availability and latency, and build one alert that fires on objective burn rather than a raw threshold
Inject a dependency failure and verify the service degrades gracefully through timeout, retry with backoff, and circuit breaking
Your checkpoint
A load-test report with a latency-versus-throughput chart, the identified ceiling, and a dashboard or alert definition tied to the stated objectives
Development Lifecycle, Testing, and Governance & Platform Architecture and System Design & Data Modeling, Storage, and Database Management
Build a CI pipeline that runs linting, unit tests, integration tests, and a build artifact on every change
Add a schema migration step to the pipeline with a rollback path and demonstrate a failed migration being reverted
Write a test plan covering unit, integration, contract, and end-to-end layers, naming what each layer catches that the others miss
Complete two timed practice sets weighted toward the heaviest domains and log every missed question with its root cause
Revisit the week-1 architecture decision record and revise it with what you now know
Your checkpoint
A green CI pipeline with migration and rollback demonstrated, plus a scored practice-exam log with root-cause notes for every miss
Milestones
Week 2: You can whiteboard the platform's layers and justify the service boundaries without notes, and a two-endpoint service runs locally with passing tests
Week 4: You can explain a query execution plan aloud, demonstrate a rejected concurrent update, and defend a storage-engine choice for three access patterns
Week 6: The full stack is recreated from infrastructure code alone, and the role matrix shows the expected allow/deny result for every role
Week 8: CI is green end to end, a failed migration rolls back cleanly, and practice scores trend upward with no repeated root cause in your miss log