Skip to content
View AlexMercedCoder's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report AlexMercedCoder

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AlexMercedCoder/README.md

Alex Merced

Head of Developer Relations at Dremio. I build and teach the open data lakehouse, and the agent infrastructure that should sit on top of it.

I am co-author of Apache Iceberg: The Definitive Guide and author of Apache Polaris: The Definitive Guide (O'Reilly) and Architecting an Apache Iceberg Lakehouse (Manning). I then ship the catalogs, libraries, harnesses, and open contracts that make that material usable.

Orlando, FL · alexmerced.com · alexmercedcoder.dev · dev@alexmerced.com


The through-line

Most lakehouse education stops at slides. Most agent tools lock identity inside one vendor loop.

I write the books, then I write the code, then I write the contracts so other tools can do the same:

  • A named agent should live as a file you can diff, review, and reuse. Not as a process that dies with the session.
  • A project should run as a graph of bounded agent loops, with briefs, budgets, success criteria, and human gates.
  • The harness you already run should stay in charge of models, tools, sandboxing, and policy.

That is the work behind Open Agent Profile and the Agentic Graph Spec. It is also why Merced AI is a broker, not another agent loop.


Start here

If you care about… Open this
Portable agent identity Open Agent Profile (OAP) 1.0
Agent workflows with gates Agentic Graph Spec (AGS) 1.0
A terminal agent that remembers MagAgent
Iceberg tables from Python IceFrame
A Rust Iceberg catalog Pangolin
Using the harnesses you already have Merced AI

Open specifications

These are the most important repositories in this profile. They are implementation-neutral, Apache-2.0, and meant to outlive any one CLI.

Open Agent Profile persists a named AI agent as data: role, model tier, tool surface, permissions, attached context, and what previous sessions learned. Three rules keep it safe to leave enabled. A profile may narrow capability and never widen it. An agent cannot rewrite its own contract. Learned state is treated as untrusted content.

Agentic Graph Spec describes a project as a DAG of bounded agentic loops. Each node is a unit of work an agent can run end to end. Each edge is a control-flow dependency. Intelligence is a portable tier, not a hardcoded model name, so a harness can route the work.


Software

Local-first agents

Project What it is
MagAgent Terminal-native coding and productivity agent with persistent memory, skills, MCP, and OAP profiles
MagGraph Rust graph memory: Markdown nodes, search, backlinks, recall bundles, Python bindings
Mag Command Center Desktop UI over the MagAgent CLI, MagGraph memory, and local workbench state
Loro Governed Python CLI harness for coding, data work, and AGS workflows
Merced AI Local broker that discovers installed harnesses and runs portable OAP bots on them

Mag is short for magpie. Corvids remember, manufacture tools, and plan ahead. That is the design brief.

Lakehouse tooling

Project What it is
Pangolin Rust lakehouse catalog: Iceberg-compatible, multi-tenant, Git-style branching, federated catalogs
IceFrame DataFrame-style Python API for Iceberg tables over REST catalogs, with local PyIceberg / PyArrow / Polars execution
Lakehouse Tools Catalog and lakehouse utilities

Pangolin and IceFrame are alpha. Use them, break them, file issues.


Books

Start with the three publisher titles:

The rest of the catalog, including the agentic analytics and AI engineering titles, lives at books.alexmerced.com.


Teaching and speaking

I educate practitioners on open lakehouse architecture, Apache Iceberg, Apache Polaris, semantic layers, and agentic analytics through Data Lakehouse Hub, Iceberg Lakehouse, LinkedIn Learning, and the Datanation and Web Dev 101 podcasts.

I speak at Data Council, Data Day Texas, Øredev, Confluent Current, Subsurface, Iceberg Summit, and similar rooms. For talks, workshops, and panels: Contact@alexmerced.com.

In 2025 I was the inaugural winner of Dremio's CEO Award.


Elsewhere

Apache Iceberg, Apache Polaris, Apache Parquet, and Apache Arrow are trademarks of the Apache Software Foundation. This profile is independent of the ASF.

Pinned Loading

  1. apache-iceberg-ui apache-iceberg-ui Public template

    basic app you can run locally to browse and query your Iceberg Catalog

    Python 12 3

  2. SencilloDB SencilloDB Public

    repo for SencilloDB a JSON based Javascript Datastore

    TypeScript 2

  3. iceberg-catalog iceberg-catalog Public

    basic python based Iceberg Catalog for testing and practice purposes, feel free to fork and extend

    Python

  4. developer-advocacy-dremio/dremio-cloud-dremioframe developer-advocacy-dremio/dremio-cloud-dremioframe Public

    Dremio Dataframe Library, DremioFrame

    Python 2 2

  5. Pangolin Pangolin Public

    Pangolin is an Open-Source MIT Licensed Data Lakehouse Catalog in RUST with Iceberg REST Catalog Support

    Rust 17 3

  6. iceframe iceframe Public

    Dataframe like library and AI Agent for working with Apache Iceberg in Python, using pyiceberg plus natively implemented procedure extensions

    Python 24 4