FP

@focalpoint

AWS Ship It Track

Intent-aware assistive foveated vision for low-vision developers watching cloud lectures

💾 github•live on aws
PRESETS:
⚡ AWS Lambda Orchestrator (Python 3.12)
🌐 Amazon API Gateway (REST API)
🗄️ Amazon DynamoDB: Pathology Profiles
👤 Dr. Werner Vogels (Lip-Reading Focus)
👁️ Amazon Rekognition: Text & Facial Mesh
🔊 Amazon Polly: Neural Audio Description
📊 Live CloudWatch Event Telemetry Stream
🌐 AWS re:Invent 2026: Masterclass
Foveal Focus ReadyDirect cursor or gaze to magnify

Hover over or gaze at any architectural component (DynamoDB, Lambda, API Gateway) to smoothly magnify and isolate it.

|Dwell: 0%|
ZOOM:
CLINICAL:
CONTRAST:

AWS Cloud Services & Telemetry Orchestrator

HTTP 200 OK

Event-driven multi-model extraction via Amazon API Gateway, AWS Lambda (Python 3.12 Graviton), DynamoDB, Rekognition & Polly

Round-Trip Latency653ms
API Gateway & Lambda
24ms dispatch

Python 3.12 Graviton orchestrator. Evaluates edge frame differentials and coordinates concurrent inference.

focalpoint-orchestrator
Amazon DynamoDB
● SYNCHRONIZED (3.4ms)

Sub-millisecond retrieval of user ophthalmic profiles: AMD central scotoma, 280ms dwell, and amber contrast preferences.

usr_kanak_0013.4ms read
Amazon Rekognition
99.2% Conf

DetectText OCR extracts semantic bounding boxes from slide code while 468-point facial mesh tracks presenter face.

8 Entities • OCR + Face Mesh
Amazon Polly
Neural Joanna

Neural text-to-speech engine synthesizes 24 kHz high-fidelity audio readouts of technical slide diagrams on demand.

24,000 Hz MP3 Audio Stream
SERVERLESS PIPELINE TELEMETRY PAYLOAD|HTTP 200 OK•Latency: 653ms
AWS Reference Pipeline Ready
{
  "status": 200,
  "processingLatencyMs": 653,
  "frameId": "frm_aws_master_001",
  "dimensions": {
    "width": 1280,
    "height": 720
  },
  "dynamoDbProfile": {
    "userId": "usr_kanak_001",
    "pathology": "AMD (Central Scotoma)",
    "readLatency": "3.4ms"
  },
  "rekognitionOCR": {
    "confidence": 0.992,
    "regionsExtracted": 8,
    "facialMesh468": true
  },
  "pollySpeech": {
    "voice": "Joanna (en-US Neural)",
    "sampleRate": "24,000 Hz",
    "status": "READY"
  }
}