AF
Amir Firouz · Portfolio
AandZ.tech · Amir Firouz

Independent Senior Software Engineer
AI Integration & Distributed Systems

Founder of AandZ.tech, an independent AI & cloud engineering practice helping teams ship reliable distributed systems and practical AI features—from LLMs, RAG, and agents to image models and feature routing in production.

I’m Amir Firouzmanesh, a senior software engineer (PhD) working where solid distributed systems meet applied AI. I’ve shipped features at Microsoft, AWS, and Property Finder—from AI-powered editing in Windows Photos to RDS SQL Server capabilities and event-driven subscription/credit platforms behind real-estate listings.

These days I focus on integrating LLMs, RAG, and agent-style workflows into existing products without turning everything into a research project. A lot of my thinking lives in open-source reference systems—like Credit Ledger, AI Feature Router, and Learning Path Designer—which I use as blueprints for client work and teaching.

AI Integration & GenAIEvent-Driven SystemsCredits & LedgersFractional Principal Engineer
Portrait of Amir Firouzmanesh
+20%
Monthly Active Users (Windows Photos)
−50%
Support Tickets (Credits Platform)
1h → 10m
Onboarding Time (DR Automation)
Previously shipped at
MicrosoftAmazon Web ServicesProperty FinderCenturyLink / DataGardens
Projects

Flagship reference implementations

Fintech Core: Credit Ledger

Production-ready reference architecture for SaaS credits, marketplace wallets, and usage-based billing.

Case study
Reference implementation of a double-entry credit ledger with CQRS, transactional outbox, and idempotent consumers—designed as a starting point for subscription/usage platforms.
GoPostgresKafka/RedpandaJaegerGrafanaDocker

Retail AI – Polyglot Search & Feature Router

A unified high-performance architecture: Go for orchestration, Python for intelligence, and an AI Feature Router for cost control.

Case study
A production-grade retail discovery engine solving "zero results" via Hybrid Search (Dense + Sparse). Features a custom AI Feature Router to cut cloud inference costs by 90% while maintaining sub-100ms latency.
QdrantFastAPIHybrid SearchOpenTelemetryPythonDocker

LearnPath – Agentic Workflow Orchestrator

A reference architecture for reliable AI agents implementing Planner-Executor patterns and Structured Outputs.

Case study
A reference architecture for reliable AI agents. Implements the Planner-Executor pattern to decompose goals into verified steps and uses Structured Outputs to prevent hallucinations—bridging the gap between Python prototypes and Go production systems.
Next.jsGoPythonQdrantPostgresDocker