Elixir/BEAM systems, Bayesian inference, and production analytics. Two decades of enterprise engineering distilled into three practice areas.
Three practice areas, one engineering discipline.
End-to-end analytics pipelines from legacy databases to real-time dashboards. Semantic layer design with Cube.dev, Apache Arrow connectors, and OLAP architectures that let the business ask questions without waiting.
From exploratory analysis to production-deployed models. Bayesian methods, time series forecasting, regime detection, and probabilistic programming for quantitative decision-making.
Production systems on the BEAM virtual machine. Fault-tolerant, distributed, real-time. From Phoenix web applications to custom OTP supervision trees and Rust NIF integrations.
Deep technical capability built over two decades of enterprise and systems engineering.
Production Elixir/Erlang since OTP 21. Supervision trees, GenServers, distributed Erlang, hot code upgrades. Custom Scenic drivers and renderers for native desktop, Android, and iOS. Systems designed for five-nines uptime.
MCMC samplers, variational inference, hierarchical models. Built eXMC from scratch: a NUTS sampler with Stan-style adaptation that beats PyMC on medium-to-large models. Regime-switching models for quantitative trading signals.
PostgreSQL, ClickHouse, time-series databases, ETL pipelines, API integrations. Cube.dev semantic layer with Ecto schema introspection. Apache Arrow ADBC drivers. From Alpaca market data streams to analytics dashboards.
Rust NIFs for compute-intensive hot paths. GPU acceleration via CUDA and Metal (EXLA, EMLX). NanoVG/OpenGL rendering pipelines. Binary protocol design for cross-platform Scenic drivers. Low-level C/C++ when the problem demands it.
AWS certified. Kubernetes orchestration, Helm charts, Ansible automation, Docker containerization. Distributed Erlang clusters on cloud infrastructure. CI/CD pipelines and infrastructure as code.
Two decades across enterprise platforms: Java/Spring ecosystems, DB2, WebSphere, Salesforce, GraphQL APIs. Legacy system modernization from mainframe-era databases to real-time Elixir pipelines.
A probabilistic programming framework built on the BEAM, where streaming, fault tolerance, and distribution are consequences of the runtime. Open source under Apache 2.0 with commercial licensing for trading applications.
Bayesian regime-switching models for quantitative trading signals. Paper trading trials with live Alpaca integration. Real-time visualization through Scenic. GPU acceleration via CUDA and Metal.
Explore the technical story →