Engineering blog

objgit is not a git forge. It's a git server on Tigris object storage that you can script with hooks, explore over SSH, and grow with. Here's what's next.

Goldsky built Prism, a geo-replicated blockchain data lake, on Tigris. TAG, the Tigris Acceleration Gateway, serves historical point lookups in about 9 ms without a database copy in every region.

FoundationDB was our only database, queue included. We built Apple's QuiCK design on top of it, hit the write limits, and moved async tasks to Kafka.

Git packfiles were designed for mmap and local disk, so pulling one object out of a bucket means guessing at a byte range. I wrote a new format instead.

Ampbase runs its control plane on Tigris object storage with no database underneath. Here are the four database primitives they built, and what broke.

How Tigris resolves write conflicts across regions: timestamp-based replication on FoundationDB, clock skew races, and the tradeoffs of each bucket type.

How Tigris composes ACID metadata, global placement, caching, replication, and background work into a multi-region object store.

Deleting data is hard in a geo-replicated active-active database. Here's how Tigris built a Recycle Bin for objects and buckets on top of immutable storage.

SigV4 looks simple: sign a request, check the signature. Then you implement canonicalization, clock skew, and a cache that isn't allowed to hold your key.

The era of manual install scripts is over. Why agent-native onboarding keeps developers in flow, what tigris init --agent prints, and how to write the prompt.

Self-hosted ClickHouse runs on cheap compute, but its backups and cold tier usually live on AWS S3, and AWS charges $0.09/GB to read your own data back. Point BACKUP TO S3 and TTL tiering at Tigris instead. You keep the same stock ClickHouse features, you pay zero egress fees, and disaster recovery…

Presigned URLs are replay attacks you commit on purpose. How SigV4 signs the clock, what a presigned URL grants on Tigris storage, and what it costs you.

Move an entire bucket to Tigris in one command. The Tigris CLI scans every object and migrates it safely from any S3-compatible provider. Here's how to set it up.

Your agent runs in a disposable sandbox, but it can't live there. A breakdown of everything in the agent's world, and why it should be one forkable bucket.

Fork a customer's entire LangGraph agent state, replay their conversations through a prompt change, and judge it head-to-head before you ship to prod.

objgit is a single-binary git server that stores repositories directly in Tigris — no disk, no git binary, no database. To my shock and horror, it worked.

Tigris bundles let you pull thousands of objects in one HTTP request as a streaming tar archive — no more one GET per object. Here's how it works and why.

Tigris now supports soft delete for buckets and objects. Enable it once, and every delete becomes recoverable for up to 90 days before it's permanently removed.

The Go Storage SDK adds first-class methods for bucket forking, snapshots, and object renaming — plus a simpler high-level client for everyday storage code.

StorageSDK is a provider-agnostic storage API for Node.js: one interface across S3, R2, Azure, GCS, Tigris, and more, with snapshots and forks built in.

Kefka is a userspace shell sandbox in Go that gives every AI agent its own copy-on-write Tigris bucket fork plus Python, jq, and ripgrep via WebAssembly.

Tigris lifecycle rules now support multiple rules per bucket with prefix filters, so you can mix transitions and expirations across different prefixes.

We embedded agent-shell on the Tigris homepage so you can ls, cat, and cd through your buckets like a normal filesystem. Here's how it works.

How Basic Memory uses per-tenant Tigris buckets, rclone bisync, and bucket snapshots to give every user a portable Markdown knowledge base.

A self-updating knowledge base on Tigris, maintained overnight by an Anthropic SDK agent loop running through @tigrisdata/agent-shell. Atomic flush on success, discard on throw. The bucket either lands a clean run or stays byte-for-byte unchanged. ~$7 a month, presigned digest URL in Slack each morn…

S2 Lite turns S3-compatible object storage into a durable streaming platform. Run it on Tigris for high-cardinality workloads — per-agent reasoning traces, event firehoses — with zero egress fees and global replication.

Meet @tigrisdata/agent-shell: a virtual bash environment that gives AI agents persistent storage on Tigris buckets, with forks and snapshots for safety.

Storage-protected Mastra agents: a reference implementation. Isolated session forks, validate before promoting, contain compromise at the storage layer.

A new TypeScript SDK giving AI agents the storage primitives they need: forks, workspaces, checkpoints, and event-driven coordination on Tigris.

Use Tigris object notifications to coordinate AI agents without deploying a message queue. A writer/watcher pattern turns your bucket into an event bus.

Three bugs, one seam. How Tigris used Antithesis under fault injection to find and fix cache coherence bugs — tombstone barriers, reorder, three-layer defense.

Bucket forking on Tigris gives every agent its own sandbox with no data duplication. One Lance file, two query engines, globally distributed reads.

Tigris now ships as a plugin for Claude Code, Cursor, and other AI coding agents. Five skills, a storage subagent, and security rules give your agent full Tigris fluency out of the box.

New Tigris accounts serve public content from a dedicated domain — t3.tigrisfiles.io — separating public content delivery from API infrastructure.

Today we're announcing CougarLLM, a globally distributed inference server that brings zero-egress inference to any open-weight model.

Tigris now lets you choose exactly how your data is replicated across regions. Pick from four bucket location types (global, multi-region, dual-region, and single-region), each with built-in consistency and availability guarantees.

Learn how Mixpeek uses Tigris Object Notifications and globally distributed object storage to build an event-driven ingestion pipeline for multimodal media.

Training ML models directly from object storage is possible with the right access patterns and client tooling. We benchmarked Tigris against AWS S3 and introduced the Tigris Acceleration Gateway (TAG), a local caching service that significantly reduces epoch duration and worker requirements for mu…

Pixeltable turns your database into a reactive data layer -- embeddings compute automatically on insert. Pair it with Tigris for globally distributed ML apps.

The Tigris Partner Integration API lets platforms offer S3-compatible object storage to their users without managing accounts, regions, or per-tenant limits.

Parseable unifies agent, LLM, and infrastructure observability in a data lake on globally distributed Tigris object storage, with 80-90% Parquet compression.

We shipped our CLI with Node.js and npm, then compiled it into a standalone binary with Bun in a matter of hours. Here is what we learned and how it performed.

Agentuity built custom cloud infrastructure for agentic workloads. Tigris powers five storage layers, from agent filesystems to sandbox snapshots.

Discover Tigris' Agent Skills Library - reusable skills that help AI agents automate image optimization, conventional commits, and Tigris SDK operations.

Tigris CLI: Manage object storage from your terminal with this UNIX-philosophy tool. Install with npm and start uploading files in seconds. Try it free today.

Tigris now has an official Go SDK! Learn how storage-go provides both an AWS S3 drop-in replacement and an ergonomic high-level interface for Go developers.

Learn how we backfilled our Discord community with existing Q&A from another forum using Tigris, ETL pipelines, and AI agents, making it searchable via Answer Overflow and MCP.

Learn how to use Tigris object storage as a powerful key-value store for both structured and unstructured data, eliminating the need for separate databases.

The S3 API won because it was an interface that stayed boring. Now it's the backbone of cloud portability, enabling workloads to run on any cloud.

Growth exploded, not just in users, but in the product itself. Tigris has matured into the storage backbone for leading AI companies.

Discover the optimal RAG chunking strategy without manual testing. This Python app uses Tigris bucket forking to run parallel experiments, automatically comparing multiple chunking strategies and ranking them using objective span-based metrics.

When building a remote MCP server, you will most likely need an OIDC layer between clients (Claude, Cursor etc) and your upstream IdP (Auth0, Clerk etc). We have created a ready to use solution for you that's vendor neutral: use any IdP you want.

Deploy a self-hostable, private Docker Registry effortlessly using a Northflank 1-Click stack combined with Tigris for secure, scalable storage and authentication. Take back control of your container image destiny, avoid public registry limits and egress fees, and set up object lifecycle rules to op…

Whole bucket snapshots, like git tags for your bucket. Snapshots enable point-in-time recovery, reproducibility, and version control for object storage - finally bringing snapshot semantics to the entire object storage namespace.

Learn how Tigris implemented the Model Context Protocol (MCP) server's challenging OAuth authentication using an OpenID Connect (OIDC) "man-in-the-middle" pattern backed by Tigris as a key-value store to manage dynamic clients.

Tigris' hosted MCP server is now globally available, simplifying integration into AI workflows like ChatGPT, Claude, and agentic coding. Benefit from no installation, OAuth security, and support for multiple Tigris organizations.

Learn how AutoMQ paired with Tigris object storage redefines Kafka scaling by eliminating replication overhead, enabling instant partition reassignment, and delivering a fully elastic, cloud-native streaming engine.

Tigris bucket forking lets you make copies of terabytes of data in microseconds thanks to making objects immutable and ordering time backwards. This gives you object-level snapshotting, forking, and strong deletion protection, revolutionizing object storage with Git-like forking for massive datasets…

Fork buckets like code. Get instant, isolated copies of large datasets for AI training, experimentation, and multi-agent workflows. Achieve reproducibility, safe experimentation, and point-in-time recovery with Tigris snapshots and forks.

How immutable, append-only architecture enables O(1) snapshots and forks in Tigris object storage, inspired by ZFS and log-structured systems.

Prevent rogue agent behavior by giving every agent an instant, isolated bucket fork—no data duplication, safe concurrency, and easy rollback with snapshots.

Fork your data like you fork your code. With Tigris, create isolated copies of your data for development, testing, or experimentation without waiting for full copies.

Want to get started training your language models using PyTorch and Tigris? Here's how to get started!

Five talks we're most excited about at PyTorch Conference 2025, showcasing innovation in AI infrastructure, storage, and performance optimization.

Led by Spark Capital, with participation from Andreessen Horowitz and others

Learn how we create AI-generated illustrations featuring Ty the tiger for our blog posts, including the creative process and the open-source Tygen app built with Go, HTMX, and OpenAI API.

We compared Flux Kontext, Nano Banana and Image GPT models. Find out which one is the best for your use case.

Introducing the Tigris Storage SDK for JavaScript and TypeScript - a simplified alternative to AWS S3 SDK with environment-based configuration, error-as-values pattern, and streamlined object storage operations.

How Tigris uses Prometheus metrics for usage-based billing instead of traditional event-driven systems, reducing complexity while ensuring transparency and customer-friendly error handling.

Get unlimited message retention and truly global operation with Bufstream and Tigris.

Discover how to efficiently train large-scale ML models using LanceDB and Tigris. This guide details streaming massive datasets directly from object storage to PyTorch, simplifying distributed training and eliminating infrastructure complexities for scalable AI.

LogSeam transforms log management with Tigris, achieving massive cost savings and performance improvements in their security data lake architecture.

Supplemental materials for the DevOpsDays Halifax 2025 talk and some extra info that didn't make the cut.

Learn how to build a text-to-video generation app using Google's Veo3 model and Tigris object storage to create high-quality video content from text prompts.

Discover why OpenAI's gpt-oss model family is ideal for building reliable and safe AI agents, not just for developers.

How AI is transforming software development from autocomplete to autonomous agents, and what this means for the future of coding.

Alibaba's new open-weights Qwen Image model shows promise in text rendering but struggles with making the text appear natural. Sometimes it just looks like it was photoshopped in because that's how they assembled the training data.

Let your datasets roam freely between the multicloud with Tigris! Today you'll learn how to import your datasets into Tigris in a snap.

Tigris introduces an IAM Policy Builder that simplifies permissions and provides a VSCode-like editor experience. This tool aims to make IAM policy development easier and more secure by focusing on least privilege from the start.

Get started integrating WarpStream with Tigris so you can have a durable multi-cloud message broker that's fast anywhere on the planet!

We have spent a lot of time optimizing Tigris for small objects. This post shares the results of our benchmarks of comparing the performance of Tigris, AWS S3, and Cloudflare R2 when storing and retrieving small objects.

Conqueror the complexity of Python environments effortlessly. Discover how Development Containers offer a consistent, reliable setup for any Python project, eliminating dependency frustrations and boosting developer productivity.

Learn how Tigris Data's tigrisfs provides a native filesystem interface for unlimited, globally accessible cloud storage. Bridge the gap between cloud and local, enabling seamless integration with existing tools and tackling big data challenges for AI, analytics, and more.

Learn how to implement data time travel with Ducklake and Tigris to easily recover from database mishaps. This guide demonstrates how Ducklake creates snapshots for every `INSERT` or `DELETE` operation, allowing you to explore your analytics database as it was before disaster struck.

Combine SQL and object storage data seamlessly with DuckLake, a data lakehouse solution that works with Tigris. Query big and small data without egress fees and simplify your data analytics workflows.

Eliminate data migration risks with Tigris's Shadow Bucket Migration. Learn how to move data seamlessly with incremental, two-way door strategies and avoid downtime. Discover the benefits of read/write fallback and multi-region synchronization for object storage.

Anubis is a Web AI Firewall Utility that helps administrators keep their sites up in the wake of abusive scrapers. In order to make Anubis filter things better, you need data. Here's how Xe got that data to improve Anubis' filtering powers.

We detected fake CAPTCHAs on our platform. Here's what we did to remove them and stay a step ahead of attackers.

Tigris has support for object lifecycle rules, allowing you to automatically move files between storage tiers. Here's the gorey details on how we did it.

We built a tool to upload a file to a bucket and then create a sharable URL. Sounds so easy an AI could do it, but can it? Models struggle to break down multi-step tasks, so we design higher order tools for them.

Tigris lets you store whatever you want including vector embeddings with LanceDB! Find out how to get started today.

Tigris lets you set any domain you want to point to your buckets instead of making the domain name match the bucket name.

Sharing buckets is a breeze when you don't have to go into advanced IAM purgatory to get it done.

Tigris lets you rename objects instead of just copying them. Just add X-Tigris-Rename!

AI tools can make coding easier, but they can also introduce subtle errors that can turn out really bad if you're not careful.

Tigris delivers S3-compatible, globally distributed object storage, enabling seamless multi-cloud workflows without tool changes. Built on S3’s API standard, it simplifies scaling, avoids vendor lock-in, and supports dynamic global data placement.

Want to do vibe coding with Tigris? We released a MCP server so you can make it happen. 😎

Tigris puts your objects all over the world, but how do we do it? Today we'll break down the strategies to Tigris' global replication and how it combines fundamental parts of distributed systems to make it happen.

Tigris has achieved SOC 2 Type II certification, signifying our high standard for security and operations.

Caught the talk at SCaLE but want to learn more? Here's a list of resources that were mentioned during the talk.

Google Container Registry's deprecation has left developers seeking alternatives. Depot's container registry built on Tigris offers a fast and reliable solution for storing and distributing Docker images globally, with faster image pulls and no rate limits.

A pull-through cache is a local cache of Docker images that can be used to speed up deployments and defend against the upcoming rate limit decreases for the Docker Hub. Learn how to make one on top of Tigris!

We're transitioning to virtual hosted style URLs for all new buckets created after February 19, 2025.

Your AI workflows rely on models that other people post on the Internet. How can you be sure that they'll stay up? Today Xe covers the history of the infamous left-pad incident of 2016 and how it could happen again with AI.

DeepSeek R1 is a frontier-grade reasoning model that you can run on your own hardware. In this article, Xe digs through the papers and slices past the hype to explain what DeepSeek R1 really gives users and how the model is revolutionary for what it is.

You can update your S3 libraries again. All is good.

AI models, comprised mainly of floating-point numbers, function by processing inputs through various components like tokenizers and embedding models. They range in size from gigabytes to terabytes, with larger parameter counts enhancing performance and nuance representation. How do they get so large…

If you've recently upgraded boto3 or the JavaScript S3 client, file uploads may fail due to a required Content-Length header. Downgrade boto3 to version 1.35.x and the JavaScript client to v3.728.0 to resolve this issue until Tigris releases an update for compatibility with new versions.

Object storage can be used as a filesystem in Kubernetes with the right setup. Here's how I set it up on my homelab cluster and the tradeoffs I made in the process.

Learn how Beam offers serverless GPUs optimized for developer productivity, across many clouds. By moving object storage to a separate managed service, Beam no longer needed to worry about it as another variable when designing for consistency across clouds.

Training AI models on Big Data can be challenging due to the need for flexible storage and compute. Tigris, a cloud-agnostic storage layer, enables decoupling of storage from compute, making it easier to manage data and model weights across clouds. Tools like SkyPilot that abstract cloud providers a…

This AI stuff is cool, but GPU inference is not needed all of the time. Most of the time your instances stay idle, which means you're just burning investor money without any real benefit. Today we'll learn how to make your compute spread between hosts nomadically, hunting deals and using Tigris to m…

Tigris now supports storage tiers! Optimize your storage cost based on access patterns with three tiers: Standard, Infrequent Access, and Archive.

Tigris now supports object notifications! Object notifications are how you receive events every time something changes in a bucket. This post is a deep dive into the implementation of object notifications in Tigris.

Tigris lets you store your objects around the world so that you beat the speed of light getting your data to your users faster. Sometimes those include Docker images so that you can spin up more GPU time without having to download or manage your models separately. Join on in and learn how you can be…

We're Tigris, a globally distributed object storage platform built for high-performance workloads. We're making our reliability metrics public! Check out our live dashboard and see how we're providing 99.99% global availability. We're pretty sure we're the first to share this type of live production…

Learn how fal.ai offers the fastest generative ai in the world by using Tigris for object storage. Finding a storage solution built for an extremely performant global deployment was essential to keeping fal.ai accessible to a broad range of developers. Tigris not only was just crazy fast, it saved…

Exploring Tigris' shadow bucket feature for fun and thumbnails. In the blog post the feature will be used in an less conventional fashion. An additional elixir service will generate thumbnails, where source files as well as generated thumbnails are stored on Tigris, while making use of the included…

There has been a lot of enthusiasm about the recent announcement by AWS about support for Conditional Writes in S3. We will dive into why this is a big deal and what it means for developers. We will also explore CAS operations and how they relate to S3 Conditional Writes feature. Finally, we will di…

Now we get to the significantly more disruptive stuff. The things that while Tigris is an S3-compatible API it also provides features that enable entirely new use-cases and push the boundaries of what you can do with object storage. Let's see if we can't set your internal constraint-solver aflame wi…

Tigris offers pretty great object storage, but the S3 API has its limits. Here I show how you can use JuiceFS to turn your bottomless Tigris storage into a FUSE mounted filesystem where we persist model data for more efficiently running production-scale Ollama on Fly.io

At Tigris we offer a number of novel and practical improvements beyond what your typical Object Storage does. We fit these within the existing common APIs or as graceful extensions when necessary. In this post we look at how you can take control of the Tigris caching mechanism if you feel the need.

Since we launched our public beta three months ago, our usage has skyrocketed, and hundreds of early adopters have picked Tigris as their storage solution. We've implemented tons of requested features and invested heavily in Tigris' performance, security, and reliability. With that, we will start bi…

Tigris is a globally distributed S3-compatible object storage solution available that is nativeliy integrated with Fly.io. In this article, we'll explore creating AI-generated images using the Stability API and storing them in a Tigris bucket for easy and fast retrieval.

Tigris is a globally distributed S3-compatible object storage solution available that can easily be hosted on Fly.io. In this article, we'll explore how Tigris fits into the existing slate of object storage options and why you might choose one over the other.

Tigris is a globally distributed object storage service that provides low latency anywhere in the world, enabling developers like you to store and access any amount of data using the S3 libraries you're already using in production. Today, we're announcing our seed round led by Andreessen Horowitz.

Tigris is a globally distributed object storage service that provides low latency anywhere in the world, enabling developers like you to store and access any amount of data using the S3 libraries you're already using in production. Today, we're launching our public beta on top of Fly.io.

Tigris is a globally available, S3 compatible object storage service. Tigris uses FoundationDB's transactional key-value interface for its underlying metadata storage. In the first post of the series we will share the details of how we have built the metadata layer on top of FoundationDB. We will…

Tigris is a globally available, S3 compatible object storage service. In the first post of the series we will share the details of how we have built the authentication and authorization support in Tigris.

Building secondary indexes for is always about balance. A balance between a system that scales and is easy to manage and an API that is intuitive and delightful for a developer to use. Recently at Tigris Data, we have been adding secondary indexes to our metadata store that runs on top of Founda…

Client-side encryption, a fortress of digital privacy, empowers users to safeguard their information with an impenetrable shield. By arming individuals with the cryptographic key to their data, this formidable approach ensures an unparalleled level of control, fortifying the sanctity of personal inf…

Tigris is a globally available, S3 compatible object storage service that uses FoundationDB for its metadata storage. This blog post describes how we perform backups on FoundationDB, and what options exist for that purpose.

Evolution of how password based authentications evolved over time and how it led to Secure Remote Password.

Tigris is a globally available, S3 compatible object storage service. Tigris uses FoundationDB's transactional key-value interface for its underlying metadata storage. In this blog post, we will describe transactions in Tigris, delving into the topics of serializable transactions, the mechanics of t…

Why Tigris leverages FoundationDB for its metadata storage layer