# FAISS Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/faiss
> Markdown URL: https://aitinkerers.org/technologies/faiss.md
> Technology record last updated: 2026-09-18T15:13:50Z
> Generated: 2026-09-22T17:53:22Z

FAISS (Facebook AI Similarity Search): The open-source Meta AI library for high-performance, billion-scale similarity search and clustering of dense vectors.

FAISS (Facebook AI Similarity Search) is the core open-source library from Meta AI for efficient similarity search and clustering of dense vectors. It is engineered for billion-scale datasets, delivering state-of-the-art performance by optimizing the memory-speed-accuracy tradeoff. The library leverages advanced indexing structures (e.g., IVF, HNSW, Product Quantization) to manage and query high-dimensional data. Crucially, it provides complete C++ and Python wrappers, with CUDA-enabled GPU implementations that deliver a significant speedup (often 5x to 10x) over CPU-only operations. This makes it the go-to tool for large-scale applications: recommendation systems, image retrieval, and anomaly detection.

- Official technology site: https://faiss.ai
- Public AI Tinkerers demos and talks: 17
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Shop Talk](https://st-louis.aitinkerers.org/talks/rsvp_HCVQpW0TgEw)

I built ShopTalk, a multimodal shopping assistant that lets users search for products using text, images, or a combination of both, delivering more human-like and relevant results. In the demo, users can upload an image, type a query, or do both, and the system uses BLIP to generate contextual captions and CLIP to embed both text and images into a shared vector space, retrieving and ranking the most relevant products through a Redis-backed similarity search with late fusion scoring.

- Event context: AI Tinkerers St. Louis: April 2026 Meetup — 2026-04-14 — St. Louis
- Public talk page: https://st-louis.aitinkerers.org/talks/rsvp_HCVQpW0TgEw

### [Failing Fast with AI: Rapid Prototyping Using Google AI Studio](https://pereira.aitinkerers.org/talks/rsvp_BLcMVGVLlTs)

This demo explores a paradigm shift in product development: failing fast to learn faster. I’ll demonstrate how Google AI Studio can be used to rapidly prototype AI-powered applications, validate ideas, and surface potential product value—without investing significant time, money, or engineering resources upfront. The focus will be on hands-on experimentation, iteration cycles, and technical workflows rather than slides or theory.

- Event context: Cambio de paradigma AITINKERERS PEREIRA 2026 — 2026-02-26 — Pereira
- Public talk page: https://pereira.aitinkerers.org/talks/rsvp_BLcMVGVLlTs

### [Reducing Hallucinations with Retrieval-Augmented Generation: A Practical Introduction](https://tiruchirappalli.aitinkerers.org/talks/rsvp_TiqxPKsC4Q0)

This talk introduces how open-source Hugging Face models can be combined with Retrieval-Augmented Generation (RAG) to produce more reliable and grounded AI responses. Through a simple live demo, I’ll show how retrieval improves factual accuracy and reduces hallucinations, then briefly discuss real-world limitations of RAG and how newer ideas like Cache-Augmented Generation (CAG) aim to address them. The session is designed to be practical, accessible, and directly applicable.

- Event context: AI Tinkerers Trichy: January Meetup & Live Demos — 2026-01-31 — Tiruchirappalli
- Public talk page: https://tiruchirappalli.aitinkerers.org/talks/rsvp_TiqxPKsC4Q0

### [Arbiter -- LLM Evaluations with automatic interaction tracking, multiple evaluators, and extensible architecture](https://sf.aitinkerers.org/talks/rsvp_mb7z70EdD84)

I'll demo Arbiter's automatic observability for AI agents using two examples that show what makes it different: you get complete visibility into LLM calls and costs with zero instrumentation. Live demo (5-10 min): Example 1: Interaction Tracking (interaction_tracking_example.py) - Run a simple evaluation (semantic similarity check) - Show the automatic interaction log that appears - every LLM call captured with: - Exact prompts and responses - Token usage (input/output) - Latency per call - Model used - Timestamp - The key point: You didn't instrument anything. This just appears automatically. Example 2: Cost Tracking (cost_tracking_example.py) - Same evaluation, now show automatic cost calculation - Breakdown by evaluator, by model, by token type - Uses llm-prices dataset (100+ models, kept up-to-date) - Show how this scales: batch evaluation cost analysis across multiple outputs - The key point: Know exactly what each evaluation costs before you scale to thousands of agent interactions Why this matters for agent builders: When you're running 10K agent interactions daily, you need to know what's happening and what it costs. Arbiter gives you both automatically. No logging infrastructure to build, no cost tracking to implement - it's just there.

- Event context: AI Tinkerers x Google Cloud SF Meetup — 2025-11-21 — San Francisco
- Public talk page: https://sf.aitinkerers.org/talks/rsvp_mb7z70EdD84

### [Build an optimized agentic RAG agent!](https://dc.aitinkerers.org/talks/rsvp_I2_1__GU9oQ)

We'll design, simulate, evaluate and optimize an Agentic RAG assistant—trained on your data and tuned to your preferences!

- Event context: AI Tinkerers x Trilligent: November Meetup — 2025-11-11 — DC
- Public talk page: https://dc.aitinkerers.org/talks/rsvp_I2_1__GU9oQ

### [Distilling Chaos: Practical RAG Systems for Real-World Decision Workflows](https://toronto.aitinkerers.org/talks/rsvp_3wot9mIgJ5E)

This session exposes the deep mechanics of scalable retrieval systems and their evolution into semi-autonomous reasoning engines through human feedback loops. It's a transparent look at our production RAG stack that bridges the gap between research papers and operational reality. Attendees will learn: How to architect RAG systems that handle millions of documents while maintaining sub-second response times Practical strategies for hybrid search that outperform pure vector or keyword approaches Real-world lessons from developing technology in the AI boom. How human-in-the-loop feedback transforms usefulness from 60% to 95%+ Critical design decisions that determine whether your RAG system becomes a force multiplier or expensive experiment

- Event context: AI Tinkerers Toronto - November 2025 Meetup at Shopify! — 2025-11-10 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_3wot9mIgJ5E

### [Más allá del RAG: Grafos universales para datos no universales](https://santiago.aitinkerers.org/talks/rsvp_Ic5kPKxkc3I)

Hoy, la mayoría de los sistemas de information retrieval usan arquitecturas tipo RAG o GraphRAG: se basan en búsqueda semántica, embeddings y modelos de lenguaje para recuperar información relevante.

- Event context: Estructura, Risa y Ecosistemas: el nuevo ADN de la IA Chilena — 2025-10-29 — Santiago
- Public talk page: https://santiago.aitinkerers.org/talks/rsvp_Ic5kPKxkc3I

### [Teaching AI to read ancient Maya Glyphs](https://montreal.aitinkerers.org/talks/rsvp_Z7tI1x8ww7I)

The ancient Maya developed one of the most beautiful, complex, and intriguing writing systems in human history. For centuries, these glyphs recorded stories, rituals, dynasties, and the deep relationship between people, time, and the cosmos. Yet this knowledge was silenced — burned, buried, and forgotten for nearly a thousand years. Even today, only a few thousand people in the world can read Maya writing. I will the development of an AI agent designed to help bridge that gap. Using computer vision and deep learning, the system can detect glyph blocks, identify individual signs, and suggest possible readings by comparing them to a vast epigraphic corpus. It is not perfect — but it is a powerful first step toward using artificial intelligence to preserve and reactivate knowledge that was nearly lost forever. In doing so, we demonstrate how AI can be more than just code — it can be a force for memory, preservation, and cultural resilience. I will be presenting a live demo.

- Event context: AI Tinkerers Montreal – Spooky October 2025 Meetup — 2025-10-21 — Montreal
- Public talk page: https://montreal.aitinkerers.org/talks/rsvp_Z7tI1x8ww7I

### [From Guitar Manual to Training Data: Automated Domain-Specific AI Dataset Generation](https://brisbane.aitinkerers.org/talks/rsvp_DOGNvsOOFE0)

Drop a guitar amp manual into this pipeline and watch it spit out hundreds of technical Q&A pairs automatically. I generate 9 different versions of each question (basic to expert level, creative to conservative) then use FAISS to build a knowledge base. The fun part is comparing the dumb base model answers with the RAG-enhanced ones that actually know what tube saturation means. The "AutoRAG" bit automatically picks the highest quality questions because manually curating thousands of Q&A pairs from multiple documents doesn't scale - you need the AI to evaluate its own training data. Plus you get to watch Llama-3 learn the difference between preamp gain and power amp distortion in real time.

- Event context: AI Tinkerers – Brisbane Inaugural Meetup · 11 September 2025 — 2025-09-11 — Brisbane
- Public talk page: https://brisbane.aitinkerers.org/talks/rsvp_DOGNvsOOFE0

### [From Zero to Agentic Hero – Building a Multi-Step AI Agent with LangGraph & Open-Source LLMs](https://nyc.aitinkerers.org/talks/rsvp_kYcD_zTMxdc)

In this live demo, I’ll walk through how I built a fully open-source, agentic AI system that can plan, decide, and act across multiple steps — without human babysitting. We’ll go deep into the wiring: LangGraph for orchestrating reasoning loops RAG with FAISS for domain-specific knowledge Function calling for triggering real-world actions Multi-modal inputs (text + documents) Cost optimization and latency tuning for real-time use No slides, no fluff — I’ll pull up the code, show the architecture, run the agent live, and break down how each moving part works.

- Event context: AI Tinkerers August Demo Day ft. Auth0 and Comet. Demo what you've built! — 2025-08-26 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_kYcD_zTMxdc

### [How evals make or break products](https://pune.aitinkerers.org/talks/rsvp_xfEQO_bbGWU)

Bosky: LLM features don’t fail because the model is “bad.” They fail because teams ship vibe checks instead of measurable evaluation. In this talk I’ll share a practical playbook I use with product teams to move from demos to durable systems. We’ll break down: • Mapping business KPIs to evals so PMs, engineering, and SMEs align on “good.” • Logging first: traces/spans as the backbone for system-level and step-level evals. • Designing evals that are meant to fail (at first) to expose blind spots instead of overfitting. • Guardrails vs evals: when to block, when to monitor, and how to decide. • Small gold sets, smart synthetic data, and programmatic prompt optimization when real data is scarce. • Cost-aware strategies: target failure slices with user analytics instead of blanket judging everything. • Multi-turn/agent workflows and tool-selection evals. • RAG reality: fix chunking and vector dimensions before chasing fancy retrieval metrics. • Shipping gates and CI/CD: make evals part of the release, not an afterthought. Format: 35–40 minutes + 10 minutes Q&A. Attendees leave with checklists, sample metrics, and templates they can use the same week.

- Event context: AI Tinkerers Pune - Meetup — 2025-08-23 — Pune
- Public talk page: https://pune.aitinkerers.org/talks/rsvp_xfEQO_bbGWU

### [Building a university RAG LLM Middleware](https://nurnberg.aitinkerers.org/talks/rsvp_pvXhURuZ9O4)

I will showcase a RAG system I built and deployed at FAU, which serves as an AI-powered assistant over internal and public university data for students and staff. The talk will cover: Technical architecture: Python FastAPI server, hybrid retrieval (BM25 + dense embeddings via OpenAI + custom weight regression), and LLM generation pipeline. Performance tuning: Techniques used to reduce latency (asynchronous calls, caching), improve retrieval quality, and handle generation accuracy challenges in a real environment. Deployment learnings: Integrating with university systems, managing security and authentication, and monitoring user queries for iterative improvement. Live Demo Plan: 1- Query the assistant live with questions like “What is the process for thesis submission?” or “What events are happening this month?”. 2- Display generation output in real-time. 3- Walk through relevant parts of the code (retriever logic, query pipeline).

- Event context: AI Tinkerers Nürnberg First Meetup - July 3rd — 2025-07-03 — Nürnberg
- Public talk page: https://nurnberg.aitinkerers.org/talks/rsvp_pvXhURuZ9O4

### [Dating on the Edge: Iris-Based Matching and On-Device Mood Inference](https://manizales.aitinkerers.org/talks/rsvp_eac55l4Od3M)

Presentaré dos prototipos construidos a partir de mi investigación doctoral sobre patrones morfológicos y dinámicos del ojo como posibles correlatos psicológicos. 1. Social App estilo Tinder basada en iris (“Iris-Tinder”) Una aplicación funcional que utiliza imágenes de iris segmentadas y procesadas con un pipeline de visión por computador basado en Transformers (Swin-UNETR) para generar embeddings oculares. Estos embeddings se usan para crear un motor de afinidad entre usuarios sin depender de descripciones textuales ni fotos convencionales. Mostraré: cómo capturo y normalizo el iris (estándar ISO/IEC 29794-6), la arquitectura del modelo, la lógica de matching basada en distancias latentes, problemas reales como iluminación, segmentación, artefactos y reducción de sobreajuste en datasets pequeños. 2. Dispositivo edge para inferencia de estado de ánimo Demostraré un prototipo físico que corre inferencia local (sin nube) para estimar variaciones emocionales a partir de microdinámicas del ojo: pupilometría, fixaciones y saccades. El dispositivo, basado en un microcontrolador + cámara NIR, ejecuta un modelo cuantizado (INT8) entrenado con eye-tracking de alta frecuencia y optimizado con técnicas de poda y distillation. Mostraré los desafíos técnicos: throughput limitado, estabilización de señal, filtrado de ruido pupilar, y mantener latencia < 40 ms. Ambos demos están centrados en implementación real, decisiones de ingeniería y problemas no resueltos, no en valor de mercado ni diapositivas.

- Event context: 🏔️ ¡Décimo Encuentro de AI Tinkerers Manizales! — 2025-01-22 — Manizales
- Public talk page: https://manizales.aitinkerers.org/talks/rsvp_eac55l4Od3M

### [Covariate Search](https://hong-kong.aitinkerers.org/talks/rsvp_bHTSKW0CfRo)

The first practical application of covariate search (with the potential of revolutionizing the search industry) by introducing a new modality. In the demo I am showcasing how we can vectorize a set of keywords semantically and perform a search on other sets (this is not currently possible with semantic search).

- Event context: AI Tinkerers - Hong Kong Meetup (December) - Inauguration — 2024-12-19 — Hong Kong
- Public talk page: https://hong-kong.aitinkerers.org/talks/rsvp_bHTSKW0CfRo

### [AI Native React compiler](https://amsterdam.aitinkerers.org/talks/rsvp_a42GWjmALJw)

I would give a talk on building a React compiler that’s optimized for interacting with LLMs. I want to highlight the opportunity that exists from creating optimizations in the toolkit that LLMs can use. I want to dive into what technical optimizations you can apply across a vertical level to achieve better performance.

- Event context: AI Tinkerers Amsterdam - November — 2024-11-12 — Amsterdam
- Public talk page: https://amsterdam.aitinkerers.org/talks/rsvp_a42GWjmALJw

### [Faster Vector Search With Centroid Clustering](https://boston.aitinkerers.org/talks/rsvp_Xs3CCMNXWXI)

Vector search is a powerful tool for semantic natural language search, but it can be computationally intensive and slow to run on some platforms. In this demo I will show how k-means clustering similar to that used in FAISS to speed up your own vector search implementation regardless of platform. This works by pre-clustering your vectors into groups with a central vector point and searching these groups based on the central vector match score with the user query.

- Event context: AI Tinkerers October Meetup at Microsoft NERD AI — 2024-10-21 — Boston
- Public talk page: https://boston.aitinkerers.org/talks/rsvp_Xs3CCMNXWXI

### [Natural Language Compiler](https://toronto.aitinkerers.org/talks/rsvp_uRG4stRzPEM)

A programming language compiler converts a document in one language to another lower-level representation, exactly capturing required semantics while linking it to external libraries or machine-dependent information. Similarly, my "natural language" compiler aims to convert a document to another lower-level representation (graphs), capturing all possible semantics while linking to external sources of data. The motivation is to form a human/machine-readable representation that facilitates explainable AI and RAG.

- Event context: AI Tinkerers Toronto - Inaugural Meetup — 2024-04-11 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_uRG4stRzPEM

## Related Technologies

- [LangChain](https://aitinkerers.org/technologies/langchain) ([Markdown](https://aitinkerers.org/technologies/langchain.md)) — 445 public demos
- [PyTorch](https://aitinkerers.org/technologies/pytorch) ([Markdown](https://aitinkerers.org/technologies/pytorch.md)) — 273 public demos
- [BART](https://aitinkerers.org/technologies/bart) ([Markdown](https://aitinkerers.org/technologies/bart.md)) — 4 public demos
- [BM25](https://aitinkerers.org/technologies/bm25) ([Markdown](https://aitinkerers.org/technologies/bm25.md)) — 5 public demos
- [DPR](https://aitinkerers.org/technologies/dpr) ([Markdown](https://aitinkerers.org/technologies/dpr.md)) — 3 public demos
- [FastAPI](https://aitinkerers.org/technologies/fastapi) ([Markdown](https://aitinkerers.org/technologies/fastapi.md)) — 181 public demos
- [Haystack](https://aitinkerers.org/technologies/haystack) ([Markdown](https://aitinkerers.org/technologies/haystack.md)) — 5 public demos
- [RAG-Sequence](https://aitinkerers.org/technologies/rag-sequence) ([Markdown](https://aitinkerers.org/technologies/rag-sequence.md)) — 3 public demos
- [RAG-Token](https://aitinkerers.org/technologies/rag-token) ([Markdown](https://aitinkerers.org/technologies/rag-token.md)) — 3 public demos
- [sentence-transformers](https://aitinkerers.org/technologies/sentence-transformers) ([Markdown](https://aitinkerers.org/technologies/sentence-transformers.md)) — 11 public demos
- [T5](https://aitinkerers.org/technologies/t5) ([Markdown](https://aitinkerers.org/technologies/t5.md)) — 5 public demos
- [Apache Lucene](https://aitinkerers.org/technologies/apache-lucene) ([Markdown](https://aitinkerers.org/technologies/apache-lucene.md)) — 2 public demos
- [Chroma](https://aitinkerers.org/technologies/chroma) ([Markdown](https://aitinkerers.org/technologies/chroma.md)) — 8 public demos
- [Elasticsearch](https://aitinkerers.org/technologies/elasticsearch) ([Markdown](https://aitinkerers.org/technologies/elasticsearch.md)) — 5 public demos
- [Embeddings](https://aitinkerers.org/technologies/embeddings) ([Markdown](https://aitinkerers.org/technologies/embeddings.md)) — 22 public demos
- [Gemini](https://aitinkerers.org/technologies/gemini) ([Markdown](https://aitinkerers.org/technologies/gemini.md)) — 188 public demos
- [GPT-3](https://aitinkerers.org/technologies/gpt-3) ([Markdown](https://aitinkerers.org/technologies/gpt-3.md)) — 191 public demos
- [GPT-4](https://aitinkerers.org/technologies/gpt-4) ([Markdown](https://aitinkerers.org/technologies/gpt-4.md)) — 529 public demos