Real AWS infrastructure. 15 live projects running for under $30/month total.
A devtool for AWS practitioners. A read-only Python scanner reads a live account and emits a normalized IR that drives an interactive network diagram and a reachability reasoner — evaluating route tables, security groups and stateless NACLs to answer "can this on-prem IP reach this service, and on which ports?" Includes an internet-exposure scan and one-click Lucid / draw.io export, and honestly models the on-prem edge that even AWS’s own Reachability Analyzer can’t trace.
Real-time meeting transcription via WebSocket streaming. Speak into your mic or capture any meeting — Comprehend analyzes sentiment live, Bedrock generates a full AI summary with action items, and the result is delivered to your phone or email via SNS and SES.
Text a photo of your fridge and get AI-generated recipes. Text a photo of a plant and get a personalized care schedule. DynamoDB Streams fire on every database change — triggering Bedrock to generate intelligent two-way SMS. Computer vision, event-driven architecture, and conversational AI in one system.
A real product spun out of HomePulse. Short-term rental cleaners text one photo per room after every turnover. Rekognition inventories every consumable, Bedrock diffs the scan against the property baseline, and S3 Object Lock archives every photo as tamper-resistant dispute evidence. Per-property pricing with built-in abuse prevention, full host dashboard, and SMS-only signup.
An AWS-native pipeline that pulls encrypted bulk files, decrypts them in memory — plaintext never touches disk — and streams millions of rows into a Parquet / Athena warehouse. One container image runs as both a Lambda and a Fargate task, so files too big for Lambda’s 15-minute ceiling route to Fargate automatically, and every load self-cleans after five minutes. The live demo decrypts a file right in your browser via OpenPGP.js, or runs simulated samples up to 25M rows.
Serverless pipeline that extracts text, tables, and form fields from any PDF or image using AWS Textract. Handles single and multi-page PDFs via async processing. Upload a document and see structured data in seconds.
Step Functions state machine that fans out to five parallel Rekognition jobs — label detection, face analysis, celebrity identification, content moderation, and OCR — then conditionally runs Comprehend NLP. Bounding box overlay draws results directly on the image.
Upload any document or paste text — Textract extracts content, Comprehend identifies every named entity, and Bedrock maps the relationships between them. Rendered as a live interactive force-directed network graph that assembles itself in real time.
Drop any business document — invoice, contract, support ticket, job application, or compliance report. A Step Functions pipeline extracts, classifies with Bedrock, evaluates business rules with thresholds and risk flags, then routes to the right team with a full audit ticket in under 8 seconds.
Drag and drop states onto a canvas, connect them visually, let Bedrock AI generate the workflow from plain English, then execute it on real AWS infrastructure. Watch data flow between Lambda functions as animated particles with comet trails, payload tooltips, choice bouncing, parallel splits, and arrival bursts — all in real time.
Train a real machine learning model in your browser using Amazon SageMaker. Pick a dataset, tune hyperparameters via sliders, watch training metrics update live as accuracy climbs. When done, run inference on new data instantly. Real SageMaker training jobs, real ML.
The exact recommendation engine that powers Amazon.com, running live on this portfolio. Browse movies, click what you like, and watch Amazon Personalize learn your taste in real time — recommendations update after every interaction. Bedrock explains why each item was recommended.
Upload any file and watch it race through all 7 S3 storage classes — Standard → Standard-IA → Glacier Instant → Glacier Flexible → Deep Archive. A live cost calculator shows savings ticking up as your file migrates. Build lifecycle policies visually and simulate 3-year cost projections.
Your entire AWS portfolio rendered as a living 3D galaxy. Every Lambda function is a glowing star, every S3 bucket a nebula, every API Gateway a portal. When someone uses any project, particles fly through the real infrastructure in real time showing the data path. Built with Three.js, secured so no sensitive resource IDs are ever exposed.
A scheduled engine that prices live Kalshi prediction markets independently and flags where the model disagrees with the crowd. EventBridge fires a Lambda every two hours that pulls open markets, asks Bedrock for a calibrated probability on each, shrinks it toward the market price by confidence to kill phantom edges, stores the divergences in DynamoDB, and texts the owner via SNS when a real edge clears the bar.