RGReza GhalavandResume
APPLIED AI ENGINEER

Good AI starts
with good systems.

I build LLM systems and agent evaluation tools.
Explore how I approach retrieval, agent governance,
and the systems that make them dependable.

Explore the work ↓3 projects. From principles to implementation.
PYTHON/RAG/AGENT GOVERNANCE/SYSTEMS THINKING
SYSTEM OVERVIEWv.01
One engineering practice.
Three connected perspectives.
CONCEPTUAL RELATIONSHIPSELECT A PROJECT ↗
01 / PROJECT EXPLORER

Find the work that fits.

Repository-based case studies · September 2026
RETRIEVAL & GENERATION

Local LLM Chat Engine

A modular chat backend that connects document retrieval to interchangeable model adapters, with API and CLI access.

Local AI experiments often couple the model, retrieval, and API. That makes it harder to change backends, reproduce state, or run without heavy optional dependencies.

PythonFastAPIscikit-learnsentence-transformersFAISSHugging FaceOpenAIpytestDocker

The repository provides ingestion, search, chat, and streaming endpoints; CLI commands; TF-IDF snapshots; and echo, OpenAI, and Hugging Face adapters. Dense retrieval is optional.

Repository ↗
AGENT GOVERNANCE

LangGraph Governance Lab

A policy evaluation service that makes agent runs inspectable through deterministic checks and actionable findings.

Agent runs can exceed budgets, use restricted tools, or miss required approvals. Their behavior needs explicit checks and findings that can be traced to individual steps.

PythonFastAPIUvicornRequestspytestDocker

Policy validation, individual and batch run evaluation, and optional LLM remediation advice are implemented. The repository includes tests, CI, Docker packaging, and operational documentation.

Repository ↗
ENGINEERING HANDBOOK

AI Systems Notes

An engineering playbook for the decisions around LLMs: retrieval, evaluation, security, and operating the system.

AI knowledge is scattered across tools and demos. Teams need a coherent set of patterns for designing, evaluating, and operating an LLM system.

PythonMkDocsMarkdownGitHub Actions

The roadmap marks the v0.1 core handbook complete. Reproducible benchmarks and deployment recipes are in progress; reference local LLM implementations remain a later milestone.

Repository ↗
02 / THE COMMON THREAD

Engineering beyond
the model.

More about my work ↗
{ }

Explicit boundaries

Separate retrieval, generation, and policy decisions so each can be understood and improved.

✓

Inspectable behavior

Make decisions visible through deterministic checks, evaluation, and clear technical documentation.

↺

Operational thinking

Consider health checks, persistence, configuration, and failure paths alongside the core feature.

More engineering work

Additional repositories

Explore additional work across enterprise automation, analytics, and system design.

ERP AI Assistant

Enterprise-oriented workflow assistant for ERP-related automation.

  • FastAPI
  • ERP
  • LLM Integration
View repository

Telecom Analytics Lab

Analytics workflows and APIs for telecom operations datasets.

  • Python
  • Analytics
  • Operations
View repository

System Design Notes

System design and architecture handbook with practical guidance.

  • Architecture
  • Documentation
  • Engineering
View repository

Supporting technical work: Python System Lab is an early-stage experimental repository examining execution timing and failure propagation in Python iteration, with reproducible traces and tests on Python 3.11 and 3.13.