FDE Collective
Take the assessment

THE CURRICULUM

One system, built layer by layer.

It starts as a single model call. By the end, it’s live in a customer’s cloud, and you own it.

  1. Build
  2. Harden
  3. Deploy
  4. Own

Module 01 of 10 · Adds Model + context

Model + contextKnowledgeAgent loopAgents + APIsEvalsGuardrailsGovernanceCustomer cloudApps · costCustomer · FDE

  1. Module 01 · Build · Adds: Model + context

    Agent Foundations

    Every failure downstream traces back to a model or context decision made here.

    Anthropic logoAnthropicOpenAI logoOpenAIHugging Face logoHugging FaceCohere logoCohereMistral AI logoMistral AI
    1. 01LLM behaviour, tokens & context windows
    2. 02Model landscape & selection framework
    3. 03Reasoning models & thinking budgets
    4. 04Reasoning & planning patterns (ReAct, CoT, Reflexion)
    5. 05System prompts & enterprise persona design
    6. 06Structured outputs, schema enforcement & function calling
    7. 07Context engineering: budgeting, compaction & pruning
    8. 08Grounding, citations & source attribution
    9. 09Prompt optimisation, versioning & rollout control
    10. 10Multilingual & regional model choices

  2. Module 02 · Build · Adds: Knowledge

    Knowledge & Retrieval

    Most enterprise pilots fail on retrieval quality, not model quality.

    Lyzr logoLyzrLlamaIndex logoLlamaIndexQdrant logoQdrantElastic logoElasticNeo4j logoNeo4j
    1. 11Ingestion, parsing & layout-aware extraction
    2. 12Embeddings selection & evaluation
    3. 13Vector databases & index tuning (HNSW)
    4. 14Hybrid search, reranking & retrieval quality
    5. 15Advanced RAG: query rewriting, routing & fusion
    6. 16Agentic RAG end to end
    7. 17Permissions-aware retrieval & document ACLs
    8. 18OCR & document intelligence at scale
    9. 19Knowledge graphs & GraphRAG
    10. 20Knowledge freshness, sync & re-indexing

  3. Module 03 · Build · Adds: Agent loop

    Agent Architecture

    The difference between a demo that impresses and a system that runs.

    Lyzr logoLyzrAnthropic logoAnthropicComposio logoComposioMem0 logoMem0Temporal logoTemporal
    1. 21The agent loop: perceive, reason, act, observe
    2. 22Autonomy levels: agent vs deterministic workflow
    3. 23Tool-calling fundamentals & schema design
    4. 24MCP servers & clients
    5. 25Custom tools, tool auth & scoped access
    6. 26Memory types & architecture (episodic, semantic, working)
    7. 27Durable workflows & durable-execution infrastructure
    8. 28Human-in-the-loop, approval gates & bounded autonomy
    9. 29Failure modes, retries & graceful degradation
    10. 30Subagents, skills & code-mode patterns

  4. Module 04 · Build · Adds: Agents + APIs

    Multi-Agent & Integration

    Enterprise value comes from agents that reach real systems, not chat windows.

    Lyzr logoLyzrCrewAI logoCrewAIGoogle logoGoogleModal logoModalBrowserbase logoBrowserbase
    1. 31Multi-agent topologies: supervisor, router, swarm
    2. 32Role-based crews, delegation & coordination
    3. 33Parallelisation, fan-out & sandboxed execution
    4. 34Agent-to-agent interoperability (A2A)
    5. 35Agent-to-human handoff design
    6. 36SaaS connector engineering & auth flows
    7. 37Agent gateway: quotas, rate limits & fallback
    8. 38Event-driven & scheduled agent triggers
    9. 39Browser automation for no-API systems
    10. 40Live web retrieval & search as an agent tool

  5. Module 05 · Harden · Adds: Evals

    Evaluation & Quality

    Evals are the language that connects business requirements to engineering.

    Braintrust logoBraintrustArize Phoenix logoArize PhoenixLangfuse logoLangfuseScale AI logoScale AILyzr logoLyzr
    1. 41Turning requirements into eval criteria
    2. 42Golden datasets, test-set design & metrics
    3. 43LLM-as-judge & rubric design
    4. 44Trajectory & component-level evaluation
    5. 45Regression testing & CI gates for agents
    6. 46Online eval, A/B testing & feature flags
    7. 47Human review, annotation ops & feedback loops
    8. 48Querying trace & eval data at scale
    9. 49Acceptance criteria & customer sign-off
    10. 50Drift, quality regression & production feedback

  6. Module 06 · Harden · Adds: Guardrails

    Safety & Security

    The gate every enterprise deployment must clear before it goes live.

    Guardrails AI logoGuardrails AINoma Security logoNoma SecurityRepello logoRepelloOkta logoOktaSailPoint logoSailPoint
    1. 51Guardrail types & content safety
    2. 52Prompt-injection & tool-poisoning defences
    3. 53Automated adversarial red-teaming
    4. 54Responsible AI, bias & fairness
    5. 55Refusal, fallback & safe-failure design
    6. 56Enterprise auth, provisioning & SSO
    7. 57Authorization (RBAC/ABAC) & multi-tenancy
    8. 58Non-human identity & agent credential lifecycle
    9. 59Supply-chain, secrets & network isolation
    10. 60AI threat modelling (OWASP LLM Top 10)

  7. Module 07 · Harden · Adds: Governance

    Data & Governance

    Agents are only as good as the data they can legally and technically reach.

    Databricks logoDatabricksSnowflake logoSnowflakeInformatica logoInformaticaCollibra logoCollibraVanta logoVanta
    1. 61Source system discovery & data mapping
    2. 62Data readiness, warehouses & lakehouses
    3. 63Structured data + agents (text-to-SQL and beyond)
    4. 64Streaming & event-driven data for agents
    5. 65PII discovery, masking & tokenisation
    6. 66Data quality, lineage & catalogue
    7. 67Data residency, sovereignty & regional deployment
    8. 68AI governance frameworks & compliance-by-design
    9. 69Model, prompt & agent registries
    10. 70Audit evidence, logging & explainability packs

  8. Module 08 · Deploy · Adds: Customer cloud

    Deployment & Operations

    You deploy into the customer's environment, under their guardrails, on their pager.

    Lyzr logoLyzrAWS logoAWSHashiCorp logoHashiCorpDatadog logoDatadogPagerDuty logoPagerDuty
    1. 71Cloud landing zones, environments & promotion
    2. 72Containers, Kubernetes & runtime
    3. 73Infrastructure-as-code for agent deployments
    4. 74CI/CD & agent lifecycle management
    5. 75Self-hosted, VPC, on-prem & air-gapped deployment
    6. 76Release strategy: canary, staged rollout & flags
    7. 77Tracing fundamentals & OpenTelemetry instrumentation
    8. 78Observability dashboards for business stakeholders
    9. 79Incident response, on-call & runbooks
    10. 80SLOs, latency budgets & reliability targets

  9. Module 09 · Deploy · Adds: Apps · cost

    Enterprise Systems & Economics

    Adoption happens where people already work, and renewal happens on unit cost.

    Salesforce logoSalesforceServiceNow logoServiceNowElevenLabs logoElevenLabsTwilio logoTwilioOpenRouter logoOpenRouter
    1. 81CRM, ERP & ITSM integration
    2. 82Collaboration surfaces & embedded agents
    3. 83Voice stack, conversation design & channels
    4. 84Voice, telephony & contact-centre integration
    5. 85Cost & latency economics of models
    6. 86Model gateway routing, fallback & cost controls
    7. 87Fast-inference serving & latency optimisation
    8. 88Fine-tuning: when, dataset prep, technique & serving
    9. 89FinOps, chargeback & cost attribution
    10. 90Scaling & infrastructure for production agents

  10. Module 10 · Own · Adds: The customer

    Discovery, Delivery & Ownership

    Everything above is buildable. This is what makes you the person who ships it.

    Celonis logoCelonisAmplitude logoAmplitudeLyzr logoLyzr
    1. 91Workflow shadowing & process observation
    2. 92Extracting the real requirement
    3. 93Use-case qualification & prioritisation
    4. 94Feasibility, data readiness & access-scoped data
    5. 95Baselines, value hypothesis & ROI modelling
    6. 96Competing platform & ecosystem literacy
    7. 97Running a customer bootcamp on live data
    8. 98Scoping to a shippable slice & demo front-ends
    9. 99Operating inside an SI/GSI delivery model
    10. 100Handover, run-state, SLAs & expansion

Ten modules. One system in production.

100 / 100 competencies