# Amazon S3 Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/amazon-s3
> Markdown URL: https://aitinkerers.org/technologies/amazon-s3.md
> Technology record last updated: 2026-02-25T10:19:40Z
> Generated: 2026-09-23T02:32:40Z

Amazon S3 (Simple Storage Service): Scalable, durable, and secure object storage for any data volume, accessible from anywhere on the web.

Amazon S3 is the industry-leading object storage service, engineered for extreme scale and performance. It delivers 99.999999999% (11 nines) of data durability by automatically storing data across multiple Availability Zones (AZs). Customers leverage S3 for core use cases: building data lakes, hosting static websites, and backing up critical application data. The service handles virtually unlimited storage, with individual objects ranging from 0 bytes up to 5 TB, and offers various storage classes (like S3 Standard, S3 Intelligent-Tiering, and S3 Glacier) to optimize cost based on access patterns.

- Official technology site: https://aws.amazon.com/s3/
- Public AI Tinkerers demos and talks: 8
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Self-Hosting Open Weight Models: What It Actually Cost Us in Production](https://san-diego.aitinkerers.org/talks/rsvp_mMKaJmBkaYE)

We ran a self-hosted open weight vision model in production for a document pipeline processing about 105,000 pages a month, then finally measured it properly against the managed alternatives. This is the result: three approaches scored field by field against human-corrected ground truth on 100 real documents, 1,759 field comparisons, with real cost per accurate field. Two findings reframed our roadmap. First, our 83 percent accuracy looked like proof the open weight model was the right call, until we excluded six fields that were database joins rather than extraction, at which point our lead over a turnkey managed service collapsed to 2 points. We had spent months crediting the model for what was really a boring integration. Second, self-hosted inference came in at 0.7x the per-token frontier cost, not the 10x saving everyone assumes, and the model layer was only about two thirds of the bill, so the swap moves less than the per-page math suggests. You will also get the trick that made all of it measurable without a labeling budget: if your pipeline has a human review step and your output bucket has versioning on, version zero versus latest version is an already-labeled error set, field by field, going back as far as your retention. We found 276,000 corrected pairs nobody had ever read as a dataset. The harness behind the numbers is open source, so you can run the same comparison on your own data. Client details and dollar amounts are anonymized, everything else is real.

- Event context: Self-hosting Models and Managing Token Spend — 2026-08-21 — San Diego
- Public talk page: https://san-diego.aitinkerers.org/talks/rsvp_mMKaJmBkaYE

### [Live Full-Stack PDF Insights Pipeline: Flask + AWS S3 + React UI](https://nyc.aitinkerers.org/talks/rsvp_u9T-10v0GWo)

I’ll walk through the end-to-end architecture of a full-stack web application that automates product insights and user recommendations using AI. The app uses a React front-end with a Flask backend API to process data pulled from AWS S3 storage, combined with embeddings and basic NLP to make personalized suggestions. This demo will focus on the technical layers, including React component rendering, Python data pipelines, secure REST API routes, and cloud orchestration via AWS.

- Event context: Auth0 x AI Tinkerers: Fall Social Demo Night 🍁 ft. Hex — 2025-11-06 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_u9T-10v0GWo

### [Building a Sovereign Multi-GPU AI Infrastructure in a European Data Center (in Less Than One Year)](https://cologne.aitinkerers.org/talks/rsvp_uYinySVHqNU)

How does a one-year-old startup build a production-ready, multi-tenant AI finetuning platform and deploy it in Europe's leading cloud provider's data center? I'll walk through the technical architecture and hard-earned lessons from manufactAI's journey to launching the IONOS AI Model Studio—a fully sovereign platform handling LoRA training and inference across GPU clusters.

- Event context: AI Tinkerers Cologne – Inaugural Meetup — 2025-11-05 — Cologne
- Public talk page: https://cologne.aitinkerers.org/talks/rsvp_uYinySVHqNU

### [Redacting Risk: Live PII Protection in AI Workflows](https://seattle.aitinkerers.org/talks/rsvp_ABp6h5u_0Yo)

I’ll demo how ClientContext provides an easy to use solution for securely processes sensitive PDFs by automatically detecting and redacting 37 types of PII before handing data off to AI models. This enables powerful AI-driven analysis and search on private documents, without ever exposing confidential information. This is a critical component when providing data to non-local AI systems. This ensures that even complex documents can be used for AI-powered analysis and search, without ever exposing private information.

- Event context: Trust & Security in AI: Demo Night w/ Auth0 — 2025-10-30 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_ABp6h5u_0Yo

### [Easy indexing of NASCAR archived footage](https://seattle.aitinkerers.org/talks/rsvp_-X1ysKvGvHA)

I'm excited to show off what I've been doing but actually, I'm looking for feedback and advice. The last time I did this I worked on Bing Image and Video search and it was a custom pipeline and mostly we just wanted to find video clips, not really search within them for specific scenes. For NASCAR I need a low-cost rapid way to index race footage for a project I'm working on. So far, I have used Amazon Rekognition and Twelve Labs. I will demo both but if someone in the group has experience using these or other off the shelf indexing tools and integrating into Media Asset Management solutions I'd love to connect.

- Event context: October Meetup - Science Fair at Foundations — 2025-10-23 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_-X1ysKvGvHA

### [AI Professional Network Builder](https://hong-kong.aitinkerers.org/talks/rsvp_7r5C0zv1pcs)

In this demo, I will go through the product itself, solution architecture, and lesson learnt from vibe coding a product, below are some background info of the demonstrated product: Business Card Scanner: Digitize and Analyze Your Professional Network with AI The Business Card Scanner is a web application that transforms physical business cards into organized digital contacts using AI-powered text extraction and analysis. It helps professionals efficiently manage their network by automatically extracting contact information, categorizing companies by industry, and providing network intelligence through interactive visualizations.

- Event context: AI Tinkerers Hong Kong with AWS Meetup — September 29, 2025 — 2025-09-29 — Hong Kong
- Public talk page: https://hong-kong.aitinkerers.org/talks/rsvp_7r5C0zv1pcs

### [Building 'mission control' for software engineering agents](https://london.aitinkerers.org/talks/rsvp_ufx_6-dsLjY)

Live demo showing how Ona orchestrates software engineering agents working in parallel on engineering tasks. I'll walk through the technical architecture of isolated and ephemeral agent/human environments - show how agents securely execute code changes and demonstrate the integration with editors, source control, secrets managers etc.

- Event context: AI Tinkerers London Meetup - 28th August 2025 — 2025-08-28 — London
- Public talk page: https://london.aitinkerers.org/talks/rsvp_ufx_6-dsLjY

### [Nixiesearch: a new Lucene-based hybrid search engine which fine-tunes to your data](https://berlin.aitinkerers.org/talks/rsvp_jtED_DXuUG8)

Building a new search engine in 2024 sounds like a stupid idea: a new vector search startup is created every week, so how can you be different from so many competitors? But in practice, you can quickly discover that putting vectors into an HNSW index is not the most challenging part of building a search application that your customers would use and like. Relevance tuning, multi-field search, facets, filters, autocomplete suggestions - the RAG-vector search crowd is still discovering all these “novel” things. In this talk, we’re going to introduce Nixiesearch, an open-source hybrid search engine focused on solving typical search problems: * Based on Lucene. You get filters, facets, autocomplete, and complex queries out of the box with decent performance. * Cloud-native and serverless. Can use S3-compatible block storage for index persistence, being able to scale to zero. * Can fine-tune the underlying embedding model on your relevance labels (if you have them) or LLM-generated synthetic labels. Nixiesearch is still in an early development stage, so your opinion on “how to do a search engine right” is really important.

- Event context: AI Tinkerers Berlin - March 21 — 2024-03-21 — Berlin
- Public talk page: https://berlin.aitinkerers.org/talks/rsvp_jtED_DXuUG8

## Related Technologies

- [Anthropic API](https://aitinkerers.org/technologies/anthropic-api) ([Markdown](https://aitinkerers.org/technologies/anthropic-api.md)) — 66 public demos
- [React](https://aitinkerers.org/technologies/react) ([Markdown](https://aitinkerers.org/technologies/react.md)) — 220 public demos
- [Amazon Aurora](https://aitinkerers.org/technologies/amazon-aurora) ([Markdown](https://aitinkerers.org/technologies/amazon-aurora.md)) — 1 public demo
- [Amazon Bedrock](https://aitinkerers.org/technologies/amazon-bedrock) ([Markdown](https://aitinkerers.org/technologies/amazon-bedrock.md)) — 20 public demos
- [Amazon Comprehend](https://aitinkerers.org/technologies/amazon-comprehend) ([Markdown](https://aitinkerers.org/technologies/amazon-comprehend.md)) — 1 public demo
- [Amazon DynamoDB](https://aitinkerers.org/technologies/amazon-dynamodb) ([Markdown](https://aitinkerers.org/technologies/amazon-dynamodb.md)) — 1 public demo
- [Amazon Rekognition](https://aitinkerers.org/technologies/amazon-rekognition) ([Markdown](https://aitinkerers.org/technologies/amazon-rekognition.md)) — 1 public demo
- [Amazon Textract](https://aitinkerers.org/technologies/amazon-textract) ([Markdown](https://aitinkerers.org/technologies/amazon-textract.md)) — 5 public demos
- [Apache-2](https://aitinkerers.org/technologies/apache-2) ([Markdown](https://aitinkerers.org/technologies/apache-2.md)) — 1 public demo
- [AWS](https://aitinkerers.org/technologies/aws) ([Markdown](https://aitinkerers.org/technologies/aws.md)) — 38 public demos
- [AWS Lambda](https://aitinkerers.org/technologies/aws-lambda) ([Markdown](https://aitinkerers.org/technologies/aws-lambda.md)) — 14 public demos
- [Cloudian HyperStore](https://aitinkerers.org/technologies/cloudian-hyperstore) ([Markdown](https://aitinkerers.org/technologies/cloudian-hyperstore.md)) — 1 public demo
- [DeepSeek API](https://aitinkerers.org/technologies/deepseek-api) ([Markdown](https://aitinkerers.org/technologies/deepseek-api.md)) — 5 public demos
- [DigitalOcean Spaces](https://aitinkerers.org/technologies/digitalocean-spaces) ([Markdown](https://aitinkerers.org/technologies/digitalocean-spaces.md)) — 1 public demo
- [Docker](https://aitinkerers.org/technologies/docker) ([Markdown](https://aitinkerers.org/technologies/docker.md)) — 147 public demos
- [FFmpeg](https://aitinkerers.org/technologies/ffmpeg) ([Markdown](https://aitinkerers.org/technologies/ffmpeg.md)) — 16 public demos
- [Flask](https://aitinkerers.org/technologies/flask) ([Markdown](https://aitinkerers.org/technologies/flask.md)) — 24 public demos
- [Go](https://aitinkerers.org/technologies/go) ([Markdown](https://aitinkerers.org/technologies/go.md)) — 25 public demos