AlexMercedData

Alex Merced's data and AI work

Talks, the Lakehouse and AI Show, books, videos, and writing on the open lakehouse, Apache Iceberg, and agentic AI.

This site collects the work. For who Alex is, his background, and his press kit, go to alexmerced.com.

Watch the showBrowse the books
Talks

Conference talks and sessions

Alex speaks on open lakehouse architecture, Apache Iceberg, Apache Polaris, semantic layers, and agentic AI. He has spoken at events including Data Council, Confluent Current, and DataEngBytes.

Speaking topics and booking details are on alexmerced.com.

Speaking topics and booking
Video series

The Lakehouse and AI Show

Short episodes, one concept each

Most episodes run under four minutes. The series starts with OLTP versus OLAP and ETL versus ELT, works through Apache Arrow, Apache Parquet, Apache Iceberg, and Apache Polaris, then moves into language models, tokens, context windows, and prompting.

Watch the episodes

The podcast

Longer conversations on data engineering, open source, and where the lakehouse is heading.

Listen on Spotify
Books

Books on Iceberg, the lakehouse, and AI

Alex has written 77 books. These are the four from O'Reilly, Manning, and Packt.

See all 77 books
Videos

Films and the video roll

Three short music films, plus the latest uploads from the Alex Merced Data and AI YouTube channel in the video roll.

My Name Is AlexA personal, music-driven introduction.
Open the StackComposable agents, open standards, portable execution.
The Iceberg Open LakehouseOpen data architecture from files to semantics.
Writing

Articles and newsletters

Alex writes about open tables, interoperable catalogs, and a semantic layer that gives people and agents the same governed definitions.

Articles

New articles publish on Data Lakehouse Hub. The blogroll lists the latest ones and lets you search them by title.

Search the blogroll

Weekly newsletters

An AI newsletter on Thursdays and an Apache lakehouse newsletter on Fridays, both free on Substack.

Subscribe on Substack

Infographics

Architecture diagrams and explainers on the lakehouse, Apache Iceberg, Apache Polaris, Arrow, Parquet, MCP, and AI.

Open the gallery

New to the lakehouse? Start with the independent architecture guide, then try hands-on tasks.

Current releases · September 2026

Open data and agentic systems

My open-source work spans the data layer agents reason over and the portable contracts and runtimes that let them work. These are the current public releases and active development lines.

Embedded datav0.8.0

SencilloDB

A dependency-free embedded JSON document store with typed schemas, hardened transactions, atomic persistence, checksummed logs, diagnostics, repair, and selective shard writes.

Published on npmExplore project ↗
Apache Icebergv0.13.0 / 0.14.0 dev

IceFrame

A DataFrame-style Iceberg library. The 0.14 development line adds MCP 2.1, complete typing, stronger async and ingestion correctness, query explain, CDC, packaging, and security hardening.

PyPI stable / developmentExplore project ↗
Lakehouse catalogv0.8.0

Pangolin

A Rust Iceberg REST catalog with multi-tenancy, branching, RBAC, OIDC, encrypted warehouse credentials, audit logging, a management UI, two CLIs, and a Python SDK.

Pre-1.0 releaseExplore project ↗
Memory-first harnessv1.4.0

MagAgent

A memory-first personal agent that remembers you across sessions in Git-backed Markdown you can review. Version 1.4.0 adds per-run memory evidence, review-gated team memory, and expiring approval grants.

Published on PyPIExplore project ↗
Desktop workspacev1.0.0

Mag Command Center

The desktop cockpit for MagAgent, with runs, approvals, graphs, and memory in one window. Packaged for Linux, macOS, and Windows with unsigned installers. Requires MagAgent 1.4.0.

First stable releaseExplore project ↗
Agent memoryv0.4.1

MagGraph

Graph-shaped agent memory stored as Markdown in Git, with search, backlinks, recall bundles, and Python bindings over a Rust core.

Pre-1.0 releaseExplore project ↗
Governed harnessv0.22.0

Loro

The governed agent harness for data and platform teams, with verified identity, tamper-evident audit, and lakehouse-native tools. Version 0.22.0 adds multi-turn context with audited compaction plus experimental OIDC sign-in, multi-user mode, a container sandbox, and run evidence export.

Published on PyPIExplore project ↗
Agent brokerv0.8.0

Merced AI

One portable agent identity across the harnesses you already use, with honest reports of what each one drops. Version 0.8.0 adds ACP client and server support, worktree-per-bot group runs, cross-harness evals, and an experimental A2A endpoint.

Published on PyPIExplore project ↗
Open specificationv1.0.5

Open Agent Profile

Portable agent identity, authority, context, memory, and state deltas, with interoperable support libraries for Python, TypeScript, Go, Rust, and Java.

OAP document 1.0Explore project ↗
Open specificationv1.0.4

Agentic Graph Specification

Portable bounded agentic work graphs with typed inputs and outputs, budgets, permissions, success criteria, planning, and shared support libraries across five languages.

AGS document 1.0Explore project ↗
Open specificationv0.2.0

Agent Approval Interchange Specification

Exact, durable, transport-neutral human approvals for agent actions. The Python support library is at 0.2.0 and adds a shared durable file store; the TypeScript, Go, Rust, and Java libraries are at 0.1.0.

AAIS document 1.0Explore project ↗
Read the data-tools field guide ↗Read the agentic-systems field guide ↗

Work with Alex