Idempotent processing
Duplicate submissions are handled deliberately instead of blindly creating repeated workflow actions.
Engineering case study
LeadFlow AI
A controlled lead-operations workflow for intake, qualification, CRM routing, scheduling and follow-up.
01
Business problem
The workflow before automation.
Inbound lead handling can become a repetitive chain of manual qualification, CRM entry, routing, scheduling and follow-up. The simulated business scenario behind LeadFlow AI explores how that workflow can be automated without losing control over duplicate submissions, failures or lead state.
- 01
Receive and inspect incoming lead information.
- 02
Determine whether the lead meets qualification criteria.
- 03
Enter or update the lead in the CRM.
- 04
Route the lead to the appropriate next step.
- 05
Coordinate scheduling or follow-up.
- 06
Track workflow status and investigate failures manually.
02
Automated solution
A controlled system around the business process.
LeadFlow AI coordinates lead intake through a structured automation pipeline. Leads move through qualification and routing logic before CRM synchronization, scheduling or follow-up actions, while reliability controls preserve consistent workflow state.
Reduce repetitive lead administration.
Standardize qualification and routing behavior.
Keep CRM state synchronized with the workflow.
Make retries, duplicate handling and workflow state visible.
03
Architecture
System architecture at a glance.
This structured view establishes the architecture narrative. Final portfolio-safe diagrams are added during the evidence pass.
- 01
Lead intake
Inbound lead data enters the controlled workflow.
- 02
Qualification
Business rules determine lead status and next action.
- 03
CRM routing
Lead state is synchronized with the target CRM workflow.
- 04
Scheduling
Qualified leads progress toward appointment or follow-up.
- 05
Control layer
Retries, duplicate protection and audit state protect execution.
04
Workflow
How information moves through the system.
01
Capture
Receive lead information through the configured intake path and normalize it for processing.
02
Qualify
Apply deterministic qualification logic before allowing downstream actions.
03
Route
Synchronize the lead with CRM state and choose the appropriate operational path.
04
Act
Coordinate appointment or follow-up actions for eligible leads.
05
Observe
Preserve workflow state so retries, duplicates and execution outcomes remain traceable.
05
Engineering
The parts that make the automation dependable.
Qualification logic
Lead decisions are represented as explicit workflow logic rather than hidden manual judgment.
CRM synchronization
Automation coordinates application state rather than treating each integration as an isolated request.
Failure recovery
Retry and status behavior is considered part of the workflow rather than an exceptional afterthought.
06
Controls
Security, reliability and human control.
01
Duplicate protection
Repeated lead events are controlled before downstream operations are repeated.
02
Explicit workflow state
Operational state remains inspectable as a lead moves through the automation.
03
Controlled actions
Qualification and routing decisions occur before downstream business actions.
04
Auditability
Important workflow decisions and outcomes are designed to remain traceable.
07
Evidence
Proof will come from the completed repository.
Core acceptance
9 / 9
Deterministic and workflow-level acceptance checks.
Live integration
7 / 7
Integration checks covering the live project path.
Current automated suite
16 / 16
Passed after a clean database rebuild and reprovisioning cycle.
9/9 core acceptance checks and 7/7 live integration checks passed in the current automated suite after a clean database rebuild and reprovisioning cycle. This result describes the tested project behavior; it does not claim complete acceptance of every requirement in the original simulated-client brief.
Database-backed operations dashboard showing lead, review, booking, and workflow-health visibility using synthetic portfolio data.
n8n orchestrates intake, qualification, routing, CRM synchronization, and post-qualification actions while FastAPI owns business rules.
Canonical lead workspace combining qualification, CRM state, communications, and an append-only workflow timeline.
Human-review queue for ambiguous or low-confidence leads, with an explicit decision and reason required before automation continues.
Dead-letter workflow error surfaced to an operator with retry state, correlation ID, and an authorized recovery action.
Synthetic lead synchronized into HubSpot as an associated contact and deal, demonstrating the CRM adapter path.
08
Implementation
Technology and current boundaries.
Technology
Next.jsTypeScriptFastAPIPythonn8nSupabase PostgreSQLSupabase AuthHubSpotResendSlackDocker
Limitations
The current build should not be described as fully production-hardened or as complete acceptance of every original brief requirement. Remaining documented areas include stronger website-webhook authentication evidence, complete dashboard acceptance coverage, final-failure Slack alert verification, burst/performance evidence, accessibility review, rate-limit and payload hardening, and retention/deletion documentation.
09
Client relevance
What this project proves.
This project demonstrates the ability to design a multi-step sales automation workflow around qualification, external integrations, reliability controls and operational traceability.
Project resources
View GitHub repositoryDemo video coming before portfolio launch.