provider-abstraction

Route inference across providers.

Reference implementation of a list-building path for GTM engineers: a plain-English campaign brief becomes a deduped, qualified, enriched, sequencer-ready contact list in seven stages, with swappable data providers.

A Swift package that gives every app a single, unified chat interface across cloud and on-device AI providers. OpenAI, Anthropic, and Apple Intelligence all share the same protocol, message model, and optional SwiftUI components — swap providers with a one-line change.

LLM Foundation for TYPO3 - Shared AI foundation: configure LLM providers once for every AI extension, with encrypted API keys and services for chat, translation, vision and embeddings.

Provider-neutral Elixir/BEAM SDK for System One semantics: typed Noul, Choice and Score decisions, prepared evaluations, provider abstraction, batching, streaming, telemetry, OTP integration, runtime controls, model catalogs and evaluation tooling. TypeSafe is the first built-in provider via typesafe_api_sdk.

AI engineering harness POC for RepliSense regulated-document workflows, showing guides, contracts, sensors, policies, traces, evaluations, and provider abstraction.

Evidence-bound execution fabric for provider-independent model capabilities. MEF R0 proves qualified provider substitution across llama.cpp and AIR.

Foundational Elixir runtime library for deterministic CLI subprocess orchestration, normalized event and payload semantics, provider profile contracts, registry plumbing, and shared support modules for the nshkr AI SDK runtime stack rearchitecture.

ChatGPT-style chat app with token streaming and a pluggable OpenAI/Anthropic/mock provider abstraction (Next.js)

Governed AI routing and execution control plane with policy enforcement, provider abstraction, auditability, and fallback orchestration.

A.I.L. (AI Intelligence Layer) is a modular platform for managing, validating, and executing AI workflows with strict architectural boundaries. It provides prompt versioning, provider abstraction, reliability controls, and observability for scalable, controlled AI integration.

Evidence-based research intelligence for AI and Management of Technology.

A runtime and provider infrastructure for self-hosted text-to-speech.

Core Elixir primitives for building reliable self-hosted inference clients, provider adapters, transport boundaries, and operational controls for private AI runtimes across local, edge, and dedicated infrastructure.

Provider-agnostic AI command center with Direct and Council workflows

Multi-tenant ERP + n8n automation with concurrency-safe inventory, transactional outbox, and in-process AI. Deployable to Fly.io or Railway.

Provider-agnostic TypeScript service for retrieving and replying to social-media comments — Fastify, PostgreSQL with row-level tenant isolation, idempotent replies, OpenAPI 3.1, spec-gated. Senior-engineer take-home.