# Matter Settle

> Matter Settle is a small AI engineering studio in India (IST). A small team of engineers designs, builds and runs data pipelines, models, the application around them, and the deployment and monitoring. Matter Settle is an AI engineering studio, not a law firm or a legal-settlement service.

The studio is new. It has no client case studies, logos or testimonials, and the site does not claim any. Prices and pilot lengths are not published: each is set in a written scope before work starts.

## What the studio builds

- [Decision systems](https://mattersettle.com/services/decision-systems/): Simulation, reinforcement learning and honest evaluation for repeated decisions where a wrong action has a real cost, delivered as code in the client's own repository and accounts.
- [Data pipelines that run unattended](https://mattersettle.com/services/data-pipelines/): Ingestion, validation and storage built to run unattended, with checkpoints, retries, quality gates, alerts and a fallback source, delivered as code in the client's own repository and accounts.
- [The application around a model](https://mattersettle.com/services/ml-applications/): The service, tests, deployment, monitoring and runbook around a working model, so it can be released, watched and rolled back, delivered in the client's own repository and accounts.
- [LLM features built around one task](https://mattersettle.com/services/llm-features/): One LLM feature built around one task, with an evaluation set and rubric written first, so every change is measured, delivered in the client's own repository and accounts.

## Terms of an engagement (as stated on the site)

- A fixed-price, fixed-scope paid pilot, agreed in writing before anything starts. Its length and price are set in that written scope.
- Your code in your repository and your accounts from the first commit.
- Something you can run every week, not a status deck.
- Walk away after the pilot and keep everything it produced.
- Direct access to the engineer doing the work.

## Pages

- [Home](https://mattersettle.com/): Matter Settle is a small AI engineering studio in India. It designs, builds and runs data pipelines, models and the systems around them.
- [Services](https://mattersettle.com/services/): The four kinds of production AI work Matter Settle takes on: decision systems, data pipelines, the application around a model, and LLM features.
- [Decision systems](https://mattersettle.com/services/decision-systems/): Decision systems from Matter Settle: reinforcement learning and simulation for repeated decisions where a wrong action has a real cost, with honest evaluation.
- [Data pipelines](https://mattersettle.com/services/data-pipelines/): Data pipelines that run unattended: ingestion, validation and storage with checkpoints, retries, quality gates and alerts, in your own accounts.
- [ML applications](https://mattersettle.com/services/ml-applications/): The application around a model: services, deployment, monitoring and the tests that keep a machine-learning system releasable, in your own repository.
- [LLM features](https://mattersettle.com/services/llm-features/): LLM features built around one task, with an evaluation set built first, so you can tell whether a change made the feature better or worse.
- [Pilot](https://mattersettle.com/pilot/): How a Matter Settle pilot works: a fixed-price, fixed-scope first project agreed in writing, with an annotated sample scope you can read before you call.
- [How we work](https://mattersettle.com/work/): How Matter Settle works: the engineering standard, something runnable every week, and the terms we start from before any work begins.
- [Guides](https://mattersettle.com/guides/): Checklists, question lists and evaluation templates from Matter Settle for putting machine learning into production. Each has a copyable artifact.
- [About](https://mattersettle.com/about/): About Matter Settle: a small AI engineering studio in India (IST). What is true about the studio today, and what it does not have yet.
- [FAQ](https://mattersettle.com/faq/): Answers to questions buyers ask Matter Settle before a first project: the paid pilot, code ownership, time zones, and when AI is the wrong tool.
- [Contact](https://mattersettle.com/contact/): Start a conversation with Matter Settle. Email the studio from your own email app and an engineer replies. India, IST (UTC+5:30).
- [Privacy](https://mattersettle.com/privacy/): Matter Settle's privacy notice: no backend, no analytics and no third-party requests. Email goes out from your own email app.

## Guides

- [From notebook to service: a production-readiness checklist](https://mattersettle.com/guides/notebook-to-service/): A production-readiness checklist for taking a model from a notebook to a running service: tests, data checks, deployment and monitoring. Copy it.
- [Ten questions to ask any AI studio before you pay, including us](https://mattersettle.com/guides/questions-before-you-pay/): Ten questions to ask any AI studio before you pay for a project, including Matter Settle, and what a useful answer looks like.
- [One-task LLM feature: build the evaluation set first](https://mattersettle.com/guides/llm-evaluation-set-first/): Building a one-task LLM feature: write the evaluation set and the rubric first, then build. Includes a template and a rubric you can copy.

## Contact

- Email: settleyourmatter@gmail.com
- Location: India, IST (UTC+5:30). The working day overlaps European working hours and US mornings.
- An engineer replies. The site has no contact form endpoint: messages go out from the visitor's own email app.