IntelFactor.ai Platform

Enterprise AI platform for manufacturing defect detection and real-time quality inspection — built to run on the factory floor with evidence, traceability, and auditable decisions.

Live Sites

App	URL	Platform
Marketing	https://intelfactor.ai	AWS Amplify
Dashboard	https://app.intelfactor.ai	AWS Amplify
API	https://api.intelfactor.ai	AWS Lambda

Key Flows (What to click)

Setup Wizard: https://app.intelfactor.ai/edge/setup Upload SOP → extract rules → configure thresholds → deploy to edge (demo-safe).
Live Stream: https://app.intelfactor.ai/edge/live View edge video stream (or demo placeholder if device unreachable).
Ask Agent: https://app.intelfactor.ai/edge/ask Ask about defect counts, evidence, and SOP traceability (best-effort explanations).

Demo mode is designed to never block the UI. If backend/device calls fail, the experience falls back gracefully to simulation.

Inspection Model Registry (IMR)

IMR is a GitOps-style versioning + deployment system for edge AI models.

When a Jetson detects a defect, auditors and QA teams need to answer: "Which exact model version made this decision?"

IMR provides:

Immutable model versions — bundle: model + policy + thresholds + runtime config
Server-computed SHA256 verification — never trust client hashes
Desired/reported state reconciliation — cloud sets desired, device reports actual
Atomic edge updates + rollback — symlink swap for zero-downtime deploys
Evidence stamps — hash metadata attached to every defect record (audit trail)

Version Lifecycle

DRAFT ──────▶ APPROVED ──────▶ ACTIVE ──────▶ RETIRED
  │              │                │
  │ Upload       │ Server         │ Set desired
  │ artifacts    │ computes       │ state for
  │ to S3        │ hashes         │ devices
  ▼              ▼                ▼
Editable     Immutable        Devices
             Locked           pull updates

Evidence Stamping (Audit Compliance)

Every defect record includes:

{
  "defect_id": "def-12345",
  "decision": "FAIL",
  "confidence": 0.92,
  "defect_type": "chip",
  "evidence_stamp": {
    "inspection_id": "blade-inspect-001",
    "version": "1.1.0",
    "manifest_hash": "0b0fc8d9...",
    "engine_hash": "ac27a1f3..."
  }
}

This enables complete traceability: any defect can be traced back to the exact model version that detected it.

Docs: docs/IMR_SMOKE_TEST.md

Project Structure

IntelFactor/
├── components/                # Marketing site React components
├── src/                       # Marketing app entry + shared UI
├── pages/                     # App routes (Vite/React)
├── backend/
│   ├── apps/
│   │   ├── api/               # FastAPI server (Lambda/container)
│   │   ├── edge/              # Jetson edge runtime (DeepStream/YOLO)
│   │   ├── web/               # Dashboard app (React)
│   │   └── stream_writer/     # Kafka → DynamoDB consumer
│   ├── packages/
│   │   ├── policy/            # Guardrails (PASS/FAIL/REVIEW)
│   │   ├── streaming/         # Kafka producer/consumer
│   │   └── imr/               # Inspection Model Registry
│   ├── prompts/               # Versioned prompts
│   └── scripts/               # Setup & automation scripts
├── infra/                     # Terraform IaC
└── docs/                      # Documentation (see docs/README.md)

Documentation

docs/README.md — documentation index
CLAUDE.md — development guidelines for AI assistants
backend/CLAUDE.md — backend-specific context
docs/DEPLOY_PRODUCTION.md — production deployment guide
docs/OAUTH_SETUP.md — Cognito / OAuth configuration
docs/CONFLUENT_SETUP.md — Confluent Kafka setup and operations
docs/KVS_LIVESTREAM_SETUP.md — KVS live streaming setup and troubleshooting
docs/IMR_SMOKE_TEST.md — IMR guide
infra/README.md — infrastructure architecture

Quick Start (Local Dev)

Marketing / App Shell

npm install
npm run dev
# http://localhost:3000

Dashboard

cd backend/apps/web
npm install
npm run dev
# http://localhost:5174

Backend API (FastAPI)

cd backend
pip install -r apps/api/requirements.txt
uvicorn apps.api.main:app --reload --port 8000
# http://localhost:8000/health

Kafka Streaming

cd backend
pip install "confluent-kafka[avro]>=2.3.0"
make kafka-smoke-test    # Verify connectivity
make run-writer          # Start consumer

Core API Endpoints

Method	Endpoint	Description
GET	/health	Health check
POST	/api/v1/sop/parse	Parse SOP + extract inspection rules
POST	/api/v1/agent/deploy	Deploy agent config to edge device
POST	/api/v1/agent/ask	Ask agent about current line / defects
POST	/api/v1/inspect	Submit image for inspection
GET	/api/v1/metrics	Inspection metrics
GET	/api/v1/models	List available models
GET	/api/v1/events	Inspection events stream
GET	/api/v1/kvs/stream/hls-url	Get KVS HLS streaming URL

Frontend demo mode falls back gracefully if these are unreachable.

Model Registry

Model	Where	Use Case
YOLO (TensorRT)	Edge	Real-time detection (15-25ms)
Multimodal LLM (Azure/Bedrock)	Cloud	Evidence explanation, SOP Q&A

Streaming Architecture

Edge (Jetson)                    Cloud
┌──────────────┐                ┌──────────────────────────────────┐
│ TensorRT     │                │  Confluent Cloud                 │
│ Inference    │──Kafka──────▶  │  ┌─────────┐    ┌─────────────┐  │
│              │                │  │ Topics  │───▶│ Flink SQL   │  │
│ SQLite       │                │  └─────────┘    └─────────────┘  │
│ Buffer       │ (offline)      │       │              │           │
└──────────────┘                │       ▼              ▼           │
                                │  ┌─────────┐    ┌─────────────┐  │
                                │  │DynamoDB │    │ Alerts      │  │
                                │  └─────────┘    └─────────────┘  │
                                └──────────────────────────────────┘

Topics:

Topic	Purpose
`if.inspection.events.v1`	Edge inspection events
`if.metrics.aggregated.v1`	Flink-computed metrics
`if.alerts.equipment.v1`	Defect spike alerts
`if.device.config.v1`	Cloud → edge config

Commands:

cd backend
make kafka-setup           # Create topics + smoke test
make kafka-smoke-test      # Verify connectivity
make kafka-healthcheck     # Full health check
make run-writer            # Start Kafka → DynamoDB consumer

KVS Live Streaming

Real-time video streaming from Jetson edge devices via AWS Kinesis Video Streams (KVS).

Jetson (GStreamer) ──▶ AWS KVS ──▶ HLS ──▶ Dashboard (HLS.js)

Endpoints:

Method	Endpoint	Description
GET	`/api/v1/kvs/stream/hls-url?stream_name=X`	Get HLS URL (query param)
GET	`/api/v1/kvs/stream/live/{stream_name}`	Get HLS URL (header-based)

Allowed streams: intelfactor-line-1, intelfactor-line-2, intelfactor-line-3, intelfactor-factory-cam

Frontend routes:

/edge/live — Full livestream page with demo fallback
/edge/live-stream — Clean KVS player view

Docs: docs/KVS_LIVESTREAM_SETUP.md

IMR API Endpoints

Method	Endpoint	Description
POST	`/api/v1/imr/inspections/{id}/versions`	Create new version (draft)
POST	`/api/v1/imr/inspections/{id}/versions/{v}/presign`	Get S3 upload URLs
POST	`/api/v1/imr/inspections/{id}/versions/{v}/approve`	Approve + compute hashes
POST	`/api/v1/imr/inspections/{id}/versions/{v}/activate`	Set as desired for devices
GET	`/api/v1/imr/devices/{id}/desired`	Get desired state + download URLs
POST	`/api/v1/imr/devices/{id}/report`	Report running state + hashes
GET	`/api/v1/imr/inspections/{id}/audit`	Full audit trail

Jetson Deployment

The IMR updater service runs as a systemd daemon on Jetson edge devices, automatically polling for and deploying new inspection model versions.

# 1. Clone repository to Jetson
git clone https://github.com/tonesgainz/intelbase.git /opt/intelfactor

# 2. Set ownership
sudo chown -R $USER:$USER /opt/intelfactor

# 3. Create environment file (see docs/IMR_SMOKE_TEST.md for values)
nano /opt/intelfactor/.env.imr

# 4. Create bundles directory
mkdir -p /opt/intelfactor/bundles

# 5. Install Python dependencies
cd /opt/intelfactor/backend
pip3 install requests

# 6. Install and start systemd service
sudo cp /opt/intelfactor/backend/scripts/intelfactor-imr.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable intelfactor-imr
sudo systemctl start intelfactor-imr

# 7. Verify
systemctl status intelfactor-imr
sudo journalctl -u intelfactor-imr -f

Bundle Structure:

/opt/intelfactor/
├── active → bundles/blade-inspect-001/1.0.0  (symlink)
├── bundles/
│   └── blade-inspect-001/
│       └── 1.0.0/
│           ├── manifest.json
│           ├── model.engine
│           ├── classes.json
│           ├── thresholds.json
│           ├── policy.yaml
│           └── runtime.json
└── .env.imr

Troubleshooting

Amplify Build Failures (npm lock mismatch)

If Amplify fails with lock file errors, regenerate the lockfile:

rm -rf node_modules package-lock.json
npm install
git add package-lock.json
git commit -m "fix: regenerate package-lock.json"
git push

If needed, add dependency overrides:

"overrides": {
  "yaml": "^2.4.2"
}

IMR Updater Issues (Jetson)

# Check service status
systemctl status intelfactor-imr
sudo journalctl -u intelfactor-imr -n 50

# Check connectivity
curl -v https://api.intelfactor.ai/health

# Manual test
cd /opt/intelfactor/backend
PYTHONPATH=/opt/intelfactor/backend python3 -m apps.edge.tools.imr_updater

License

Proprietary — IntelFactor AI Ltd. See LICENSE for details.

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
.github		.github
amplify		amplify
backend		backend
components		components
confluent		confluent
confluentdocs		confluentdocs
docs		docs
hooks		hooks
infra		infra
jetson		jetson
public		public
scripts		scripts
services		services
src		src
tools/local-analyzer		tools/local-analyzer
utils		utils
.cursorrules		.cursorrules
.gitignore		.gitignore
.pre-commit-config.yaml		.pre-commit-config.yaml
App.tsx		App.tsx
CLAUDE.md		CLAUDE.md
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
amplify.yml		amplify.yml
index.css		index.css
index.html		index.html
index.tsx		index.tsx
og-image.png		og-image.png
package-lock.json		package-lock.json
package.json		package.json
postcss.config.js		postcss.config.js
tsconfig.json		tsconfig.json
types.ts		types.ts
vite.config.ts		vite.config.ts

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

IntelFactor.ai Platform

Live Sites

Key Flows (What to click)

Inspection Model Registry (IMR)

Version Lifecycle

Evidence Stamping (Audit Compliance)

Project Structure

Documentation

Quick Start (Local Dev)

Marketing / App Shell

Dashboard

Backend API (FastAPI)

Kafka Streaming

Core API Endpoints

Model Registry

Streaming Architecture

KVS Live Streaming

IMR API Endpoints

Jetson Deployment

Troubleshooting

Amplify Build Failures (npm lock mismatch)

IMR Updater Issues (Jetson)

License

About

Uh oh!

Releases

Packages

Uh oh!

Contributors 2

Uh oh!

Languages

License

tonesgainz/intelfactor-public

Folders and files

Latest commit

History

Repository files navigation

IntelFactor.ai Platform

Live Sites

Key Flows (What to click)

Inspection Model Registry (IMR)

Version Lifecycle

Evidence Stamping (Audit Compliance)

Project Structure

Documentation

Quick Start (Local Dev)

Marketing / App Shell

Dashboard

Backend API (FastAPI)

Kafka Streaming

Core API Endpoints

Model Registry

Streaming Architecture

KVS Live Streaming

IMR API Endpoints

Jetson Deployment

Troubleshooting

Amplify Build Failures (npm lock mismatch)

IMR Updater Issues (Jetson)

License

About

Resources

License

Code of conduct

Security policy

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors 2

Uh oh!

Languages

Packages