Developer Track

Audience: ML engineers, backend developers, AI platform engineers
Goal: Configure the VeriProof SDK in your AI application and see governed data in the portal
Estimated time: ~3 hours across 7 modules

Who can use this

Available to

DeveloperAdministrator

Not available to

Governance EngineerCompliance OfficerBusiness OwnerAuditor

This track is written for the Developer role. Administrators may also work through this track to understand SDK integration before managing production approvals.

Track overview

Self-assessment: At the end of each module there is a short self-assessment checklist. Complete all 7 checklists before requesting your completion certificate.

Module	Title	Time
1	What is VeriProof and why OTel-native?	20 min
2	Install and configure the SDK	30 min
3	Session classification and governance attributes	30 min
4	Verifying your data in the portal	20 min
5	Multi-adapter setup for complex pipelines	30 min
6	Troubleshooting common issues	20 min
7	Advanced: Custom attestation payloads	30 min

Module 1 — What is VeriProof and why OTel-native?

Goal: Understand what VeriProof captures, why OpenTelemetry is the transport layer, and what “governance-enriched span” means in practice.

Read:

Self-assessment:

I can explain the difference between an observability span and a VeriProof governance-enriched span
I understand what GovernanceInferredMask means and why it matters for auditors
I know what data VeriProof captures by default vs. what requires content capture to be enabled

Module 2 — Install and configure the SDK

Goal: Get the SDK installed, configured, and exporting spans to the VeriProof Ingest API.

Read:

Quickstart
First integration
SDK feature matrix — find the right package for your framework

Choose your language:

Self-assessment:

SDK installed and API key configured
Application ID set correctly
First spans appearing in the portal (Applications page)

Module 3 — Session classification and governance attributes

Goal: Understand the governance attribute schema and correctly classify sessions in your application.

Read:

Key concepts:

SessionIntent — advisory / decision / verification / review and others; declares the session’s purpose
RiskLevel — low / medium / high / critical; set on the session or individual steps
HumanOversightType — signals whether human review was required, granted, or applied

Self-assessment:

My application correctly sets SessionIntent on every session
I understand when to use SessionIntent.Decision vs. SessionIntent.Advisory
I understand the content capture risk profile and have made a documented decision

Module 4 — Verifying your data in the portal

Goal: Validate that your integration is working by reading the portal surfaces your data populates.

Read:

Portal checklist:

Applications page shows my application with correct session count
SDK Health page shows green for schema coverage and attribute completeness
Sessions page shows individual sessions with correct intent and governance attributes

Module 5 — Multi-adapter setup for complex pipelines

Goal: Handle pipelines with multiple frameworks (e.g., orchestration layer + LLM calls + retrieval + tool calls).

Read:

Self-assessment:

I understand how child spans attach to the parent session span
Each step in my pipeline has the correct StepType attribute
Cross-framework pipelines produce a coherent Time Machine trace

Module 6 — Troubleshooting common issues

Goal: Diagnose and fix the most common integration problems before they reach production.

Read:

Common issues covered:

Missing spans (wrong export endpoint, wrong API key)
Session intent or risk level showing as “inferred” when it should be explicitly declared
Content appearing unredacted when PII redaction is enabled
Sandbox mode leaking into production exports

Self-assessment:

I can read the SDK Health page and identify schema coverage gaps
I know what to check when spans stop appearing
I have tested sandbox mode and confirmed it does not export to production

Module 7 — Advanced: Custom attestation payloads

Goal: Understand how to extend the governance attribute schema for domain-specific audit evidence (e.g., loan decision rationale, clinical recommendation basis).

Read:

Self-assessment:

I understand which attributes are ingest-indexed vs. stored as blob
My domain-specific attributes follow the veriproof.custom.* namespace convention
Custom attributes appear in the Time Machine session detail view

Request your certificate

Once you have completed all 7 module self-assessments, request your Developer Track completion certificate. The certificate is valid for 12 months and can be presented as staff training evidence in AI governance audits.