Data foundations and AI systems that hold up in production.
Pine Marten is the independent practice of John Houghton, a senior data engineer with a decade of work at Meta, Apple, Kaiser Permanente and Intuit. I design pipelines, analytics and AI systems — including private, on-prem AI that keeps your data in your building.
What I help with
Data engineering
Modern pipelines and warehouses: ingestion, dbt modeling, orchestration, testing and CI — built to be owned by your team.
dbtDagsterDuckDBSQLAI engineering
LLM features, agents and evaluation harnesses wired into real workflows, with dataset curation and guardrails from day one.
LLMsagentsevalsLocal & private AI
On-device speech, vision and language models on Apple Silicon or your own hardware — no data leaves the premises.
on-premSTT/TTSvisionAnalytics & decision support
Metrics, experiments and dashboards grounded in economics — so the numbers answer the questions leadership actually asks.
metricsexperimentationAsynchronous quantum cryptography systems
Encryption that is simultaneously secure and compromised until someone opens the logs. Delivered asynchronously, which is to say eventually.
post-quantumpre-revenueidempotent decoherenceSmall, senior, hands-on
Assess
A short, fixed-scope review of your data and goals, ending in a written plan with clear priorities.
Build
I work in your repos and tools, shipping in small increments with tests and documentation.
Hand off
Runbooks, walkthroughs and a team that can extend the system without me.
Hi, I'm John.
I've spent my career turning messy data into dependable systems — most recently supporting FAIR and Reality Labs at Meta, and before that Apple Maps, Kaiser Permanente and Intuit. I trained as an economist, so I care about the decision the data serves, not just the pipeline.
Senior data and AI engineer in San Diego. Economist by training, builder by habit. Available for consulting engagements and select full-time roles.
Things I've built
SD Open Data Pipeline
End-to-end pipeline for San Diego open data: Dagster orchestration, dbt models, DuckDB spatial joins to council districts, and CI with full test coverage.
DagsterdbtDuckDB spatialCIsd-open-data-pipeline
LLM Dataset Curation
Config-driven curation for language-model datasets: normalization, exact and MinHash near-dedupe, Gopher/C4 quality filters, PII redaction, toxicity filtering and an eval report.
LLMsdata qualityllm-dataset-curation
Vision Lab · Local AI Showcase
Real-time camera vision and voice agents running fully on-device on a Mac Studio M3 Ultra, with event recording, data-quality checks and privacy redaction.
on-devicevisionvoicevision-lab
Finance Wizard
An agent-first personal finance app: FastAPI and React with interactive ECharts models, scenarios and dashboards.
FastAPIReactagentsfinance-wizard-demo
Personal Finance Lake
A local lakehouse on DuckDB and dbt that ingests bank feeds, enriches transactions and reconciles them nightly.
DuckDBdbtlaunchdfinance-lake-demo
Multi-Agent Assistant
A team of cooperating AI agents for research, scheduling, note-keeping and security review, with human approval on every outbound action.
agentsorchestrationmulti-agent-assistant
Distributed Multimedia ETL Pipeline
Distributed image ETL on LAION-400M samples: Spark hashing and metadata extraction, exact and perceptual near-dedupe, WebP compression, and Delta Lake medallion tables orchestrated by Airflow.
SparkDelta LakeAirflowDistributed-Multimedia-ETL-Pipeline
Multi-Provider LangChain
FastAPI and LangChain backend with a lightweight web UI for switching between Anthropic and xAI models at runtime, with chat, canvas and real-time Grok voice modes.
LangChainFastAPIvoicemulti-provider-langchain
Where I've worked
Education & training
B.A. Economics
San Diego State University, 2011
Machine Learning Specialization
University of Washington, 2017
Data Mining Certificate
UC San Diego Extension, 2017
Outside the terminal
Local-first and private AI, economics and markets, endurance sports, and rare books.
Have a data or AI problem worth solving?
Send a few lines about what you're working on. I reply within one business day.