Backend developer specializing in microservices architecture, distributed systems, and AI integration
About • Tech Stack • AI Workflow • Projects • Stats • Contact
Backend developer with expertise in building scalable, production-ready systems. Currently focused on AI-powered applications using modern architectures including multi-agent systems, event-driven patterns, and async/await paradigms.
🔧 View Complete Tech Stack
graph TB
subgraph Backend["Backend Technologies"]
Python["Python 3.14<br/>Async/Await"]
FastAPI["FastAPI<br/>High-Performance"]
CSharp["C# .NET 9<br/>Advanced Patterns"]
TS["TypeScript<br/>Type-Safe"]
end
subgraph Frontend["Frontend Technologies"]
React["React 18<br/>Hooks & Context"]
Vite["Vite<br/>Fast Build"]
Shadcn["shadcn/ui<br/>Components"]
end
subgraph AI["AI Stack"]
LangGraph["LangGraph<br/>Multi-Agent"]
GPTOSS["OpenAI GPT-OSS<br/>120B & 20B"]
Claude45["Claude Sonnet 4.5<br/>Haiku 4.5<br/>Opus 4.1"]
Grok["Grok 2<br/>X AI"]
Gemini3["Gemini 3 Pro<br/>3 Pro Image"]
Ollama["Ollama<br/>DeepSeek-R1"]
Dictalm["DictaLM<br/>Hebrew"]
Qwen3["Qwen3<br/>30B-A3B"]
Vector["Vector DBs<br/>RAG"]
end
subgraph Data["Data & Storage"]
PG["PostgreSQL<br/>pgvector"]
Redis["Redis<br/>Cache"]
MongoDB["MongoDB"]
end
subgraph Infra["Infrastructure"]
Docker["Docker"]
K8s["Kubernetes"]
CICD["CI/CD"]
end
Backend --> Frontend
Backend --> AI
Backend --> Data
Backend --> Infra
style Backend fill:#1e40af
style Frontend fill:#059669
style AI fill:#7c3aed
style Data fill:#ea580c
style Infra fill:#0891b2
Agent-based development: All development tasks are orchestrated through AI agents, leveraging MCP (Model Context Protocol) for seamless integration with codebases, external services, and custom tools. This approach enables rapid iteration, intelligent code generation, and comprehensive architecture design.
|
|
|
|
|
🔧 View Agent-Based Development Workflow
graph TB
subgraph Tools["AI Development Tools"]
Perplexity["Perplexity<br/>Research & Learning"]
ClaudeCode["Claude Code<br/>Code Generation"]
ChatGPT["ChatGPT<br/>Problem Solving"]
Cursor["Cursor AI<br/>IDE Integration"]
end
subgraph MCP["Model Context Protocol"]
MCPServers["MCP Servers<br/>Custom Tools"]
MCPResources["MCP Resources<br/>Codebase Access"]
MCPIntegrations["MCP Integrations<br/>External Services"]
end
subgraph Agents["Agent Orchestration"]
Planning["Task Planning<br/>& Architecture"]
CodeGen["Code Generation<br/>& Refactoring"]
Review["Code Review<br/>& Optimization"]
Docs["Documentation<br/>& Design"]
end
subgraph Output["Development Output"]
Code["Production Code"]
Architecture["System Architecture"]
Decisions["Technical Decisions"]
end
Tools -->|Query| MCP
MCP -->|Assist| Agents
Agents -->|Generate| Output
Output -.->|Feedback| Tools
style Tools fill:#1e40af
style MCP fill:#7c3aed
style Agents fill:#059669
style Output fill:#ea580c
Workflow: AI Tools → MCP Integration → Agent Orchestration → Development Output
Tech Stack: Python 3.14 (62%) + FastAPI | TypeScript (36%) + React 18 + Vite | shadcn/ui | Docker
|
|
|
|
|
|
🎯 The Challenge • Manual content review is slow and error-prone • Large document volumes overwhelm teams • Quality standards vary across reviewers
💡 The Solution AI-powered platform automates accuracy verification and quality assurance.
✨ Key Features ✅ Real-time content analysis and verification ✅ Multi-language support (Hebrew + English) ✅ Automated quality scoring system ✅ Semantic similarity clustering ✅ 7 microservices architecture ✅ Hybrid AI (Cloud + Local LLMs)
📊 Impact
📐 View Architecture Diagram
graph TB
subgraph Frontend["Frontend Layer"]
React["React 18 + TS"]
Vite["Vite"]
UI["shadcn/ui"]
end
subgraph Gateway["API Gateway"]
APIGateway["FastAPI Gateway<br/>Routing & Auth"]
end
subgraph Microservices["Services"]
Content["Content Service"]
Analytics["Analytics Service"]
ML["ML Service"]
Notify["Notification Service"]
Morph["Morphological Service"]
NER["PyTorch NER Service"]
end
subgraph Data["Data Layer"]
PG["PostgreSQL<br/>pgvector"]
Redis["Redis Cache"]
end
subgraph AI["AI Infrastructure"]
CloudLLM["Cloud LLMs<br/>GPT-OSS-120B<br/>Claude Sonnet 4.5<br/>Grok 2<br/>Gemini 3 Pro"]
LocalLLM["Local LLMs<br/>DictaLM (Hebrew)<br/>Llama 3.3-70B<br/>DeepSeek-R1"]
end
subgraph Monitor["Observability"]
Jaeger["Jaeger"]
Prometheus["Prometheus"]
Grafana["Grafana"]
end
Frontend -->|HTTPS/WebSocket| Gateway
Gateway -->|Route| Microservices
Microservices -->|Store| Data
ML -->|Inference| AI
Microservices -->|Traces| Monitor
style Frontend fill:#1e40af
style Gateway fill:#059669
style Microservices fill:#7c3aed
style Data fill:#ea580c
style AI fill:#dc2626
style Monitor fill:#14b8a6
Architecture: Microservices (7 services) | API Gateway | Hybrid AI (Cloud + Local LLMs) | Observability Stack
Tech Stack: LangGraph 7-Agent System | Python 3.14 (52%) + FastAPI | TypeScript (46%) + Next.js 15 + React 19 | PostgreSQL + pgvector | Redis | LLM Pool (Local + Cloud)
|
|
|
|
|
|
|
🎯 The Challenge • Market analysis requires hours of manual work • Multiple data sources are hard to synthesize • Human error affects trading decisions
💡 The Solution 7-agent AI system analyzes markets and generates recommendations in real-time.
✨ Key Features ✅ LangGraph multi-agent orchestration ✅ Real-time market data processing ✅ Sentiment analysis from news and social media ✅ Technical and fundamental analysis ✅ Risk assessment and position sizing ✅ LLM pool with automatic failover ✅ Real-time WebSocket updates
📊 Impact
📐 View Architecture Diagram
graph TB
subgraph Frontend["Frontend"]
NextJS["Next.js 15 + React 19"]
WebSocket["WebSocket Client"]
end
subgraph API["API Layer"]
FastAPI["FastAPI Server<br/>Python 3.14"]
Endpoints["REST + SSE Endpoints"]
end
subgraph Services["Services"]
MarketService["Market Data"]
SentimentService["Sentiment Analysis"]
PortfolioService["Portfolio"]
TradingService["Trading Signals"]
end
subgraph Agents["Agent System"]
Supervisor["Supervisor Agent"]
MarketAgent["Market Data Agent"]
TechnicalAgent["Technical Agent"]
FundamentalAgent["Fundamental Agent"]
SentimentAgent["Sentiment Agent"]
RiskAgent["Risk Agent"]
DecisionAgent["Master Decision Agent"]
end
subgraph LLM["LLM Pool"]
LocalLLM["Local DeepSeek<br/>DeepSeek-R1"]
GPTOSS["OpenAI GPT-OSS<br/>120B & 20B"]
Claude45["Claude Sonnet 4.5<br/>Haiku 4.5"]
Grok["Grok 2<br/>X AI"]
Gemini3["Gemini 3 Pro<br/>2.5 Pro"]
Router["LLM Router<br/>Failover"]
end
subgraph Data["Data Layer"]
PostgreSQL["PostgreSQL 16<br/>pgvector"]
Redis["Redis Cache"]
end
subgraph Observability["Observability"]
LangSmith["LangSmith"]
Prometheus["Prometheus"]
end
Frontend --> API
API --> Services
API --> Agents
Services --> Data
Agents --> Services
Agents --> LLM
LLM --> Router
Router --> LocalLLM
Router --> GPTOSS
Router --> Claude45
Router --> Grok
Router --> Gemini3
Agents --> Data
Agents --> Observability
style Frontend fill:#1e40af
style API fill:#059669
style Services fill:#7c3aed
style Agents fill:#dc2626
style LLM fill:#0891b2
style Data fill:#6366f1
style Observability fill:#1C1C1E
Architecture: LangGraph 7-Agent Orchestration | Python 3.14 Free-Threaded | LLM Pool with Failover (Local → Cloud) | Real-time Market Data | WebSocket Streaming | Correlation Discovery | Anomaly Detection | LangSmith Observability
Tech Stack: Python (FastAPI) + LangGraph v1.0 | TypeScript + React 19 | PostgreSQL + pgvector | Jina AI | SSE Streaming
|
|
|
|
|
|
🎯 The Challenge • Research and planning take weeks before coding • Technology comparisons are time-consuming • Implementation plans lack detail
💡 The Solution Automated pipeline transforms research into actionable implementation guides.
✨ Key Features ✅ Multi-agent analysis workflow ✅ Technology comparison engine ✅ Security and performance auditing ✅ Code quality assessment ✅ Real-time SSE progress streaming ✅ Automated artifact generation ✅ 7 specialist AI agents
📊 Impact
📐 View Architecture Diagram
graph TB
subgraph Frontend["Frontend Layer - React 19"]
React["React 19 + TypeScript"]
Router["React Router"]
Query["TanStack Query"]
SSE["useSSE Hook<br/>Real-time Progress"]
Components["UI Components"]
end
subgraph API["API Layer - FastAPI"]
FastAPI["FastAPI Server"]
AnalyzeEndpoint["POST /api/v1/analyze"]
SSEEndpoint["GET /api/v1/analyze/{id}/stream"]
DownloadEndpoint["GET /api/v1/artifacts/{id}/download"]
end
subgraph Workflow["Workflow"]
EntryPoint["entrypoint<br/>Checkpointer"]
Extract["extract_content<br/>Jina AI Reader"]
Embed["generate_embedding<br/>OpenAI"]
Supervisor["supervisor_route<br/>Agent Router"]
subgraph SubAgents["Specialist Agents"]
TechComp["Tech Comparator"]
Security["Security Auditor"]
ImplPlan["Implementation Planner"]
Perf["Performance Auditor"]
CodeQual["Code Quality"]
Trends["Trends Analyzer"]
Deps["Dependencies Analyzer"]
end
Aggregate["aggregate_findings"]
Artifact["generate_artifact<br/>Markdown"]
end
subgraph Data["Data Layer"]
PostgreSQL["PostgreSQL<br/>pgvector"]
AnalysisModel["Analysis Records"]
FindingsModel["Agent Findings"]
ArtifactModel["Generated Artifacts"]
end
subgraph External["External Services"]
JinaAI["Jina AI Reader<br/>Content Extraction"]
OpenAI["OpenAI API<br/>GPT-OSS-120B"]
Claude["Claude Sonnet 4.5<br/>Haiku 4.5"]
Grok["Grok 2<br/>X AI"]
Gemini["Gemini 3 Pro<br/>Image Gen"]
Ollama["Ollama LLM<br/>DeepSeek-R1 (685B)"]
end
subgraph Events["Event Broadcasting"]
Broadcaster["Event Broadcaster"]
SSEStream["SSE Stream<br/>Progress Events"]
end
React -->|HTTP| AnalyzeEndpoint
React -->|SSE| SSEEndpoint
React -->|HTTP| DownloadEndpoint
AnalyzeEndpoint -->|Create| AnalysisModel
AnalyzeEndpoint -->|Start| EntryPoint
EntryPoint --> Extract
Extract -->|Extract| JinaAI
Extract -->|Progress| Broadcaster
Extract --> Embed
Embed -->|Generate| OpenAI
Embed -->|Store| PostgreSQL
Embed --> Supervisor
Supervisor -->|Route| TechComp
Supervisor -->|Route| Security
Supervisor -->|Route| ImplPlan
Supervisor -->|Route| Perf
Supervisor -->|Route| CodeQual
Supervisor -->|Route| Trends
Supervisor -->|Route| Deps
TechComp -->|Findings| FindingsModel
Security -->|Findings| FindingsModel
ImplPlan -->|Findings| FindingsModel
Perf -->|Findings| FindingsModel
CodeQual -->|Findings| FindingsModel
Trends -->|Findings| FindingsModel
Deps -->|Findings| FindingsModel
TechComp --> Aggregate
Security --> Aggregate
ImplPlan --> Aggregate
Perf --> Aggregate
CodeQual --> Aggregate
Trends --> Aggregate
Deps --> Aggregate
Aggregate --> Artifact
Artifact -->|Save| ArtifactModel
Artifact -->|Complete| Broadcaster
Broadcaster --> SSEStream
SSEStream -->|Events| SSEEndpoint
Extract -->|Progress| Broadcaster
Supervisor -->|Progress| Broadcaster
TechComp -->|Progress| Broadcaster
Security -->|Progress| Broadcaster
style Frontend fill:#1e40af
style API fill:#059669
style Workflow fill:#7c3aed
style SubAgents fill:#dc2626
style Data fill:#ea580c
style External fill:#0891b2
style Events fill:#14b8a6
Architecture: LangGraph v1.0 Multi-Agent System | SSE Real-time Progress | Content Extraction Pipeline | Research-to-Implementation Workflow
Tech Stack: Python (57%) + FastAPI | TypeScript (43%) + React 19 | PostgreSQL + pgvector | LangGraph Multi-Agent | Redis | QStash | LangFuse
|
|
|
|
|
|
|
🎯 The Challenge • Hundreds of documents in multiple languages • Finding information takes too long • Tenant questions require manual document review
💡 The Solution AI-powered platform extracts, translates, and answers questions from property documents.
✨ Key Features ✅ Multi-language document processing ✅ On-demand translation service ✅ Natural language Q&A with RAG ✅ Semantic and full-text search ✅ Multi-agent specialist system ✅ Automated entity extraction ✅ Real-time chat interface
📊 Impact
📐 View Architecture Diagram
graph TB
subgraph Frontend["Frontend - React 19 + Vite"]
React["React 19 + TypeScript"]
Router["React Router"]
Query["TanStack Query"]
UI["shadcn/ui Components"]
Flow["React Flow<br/>Graph Visualization"]
end
subgraph API["API Layer - FastAPI"]
FastAPI["FastAPI Server"]
DocumentsAPI["/api/v1/documents"]
ChatAPI["/api/v1/chat"]
SearchAPI["/api/v1/search"]
PropertyAPI["/api/v1/property"]
FinancialAPI["/api/v1/financial"]
MaintenanceAPI["/api/v1/maintenance"]
end
subgraph Services["Service Layer"]
DocumentService["Document Service<br/>Upload & Processing"]
TranslationService["Translation Service<br/>On-Demand"]
EmbeddingService["Embedding Service<br/>OpenAI"]
SearchService["Search Service<br/>Semantic + Full-Text"]
ChatService["Chat Service<br/>RAG"]
end
subgraph Agents["Multi-Agent System"]
Supervisor["Supervisor Agent<br/>Claude 4 Opus<br/>Routing"]
subgraph SpecialistAgents["Specialist Agents"]
DocumentAgent["Document Agent<br/>Claude 3.7 Sonnet<br/>General Q&A"]
FinancialAgent["Financial Agent<br/>Claude 3.7 Sonnet<br/>Transactions & Budgets"]
MaintenanceAgent["Maintenance Agent<br/>Claude 3.7 Sonnet<br/>Work Orders"]
PropertyAgent["Property Agent<br/>Claude 3.7 Sonnet<br/>Leases & Contacts"]
end
Checkpointer["PostgresSaver<br/>State Persistence"]
end
subgraph Data["Data Layer"]
PostgreSQL["PostgreSQL 16<br/>pgvector Extension"]
DocumentsTable["documents<br/>content + embeddings"]
ConversationsTable["conversations<br/>Chat History"]
CheckpointsTable["checkpoints<br/>LangGraph State"]
Redis["Redis<br/>Cache + Rate Limiting"]
end
subgraph Background["Background Processing"]
QStash["QStash<br/>Durable Jobs"]
BackgroundTasks["Document Processing<br/>Embedding + Entities"]
end
subgraph Observability["Observability"]
LangFuse["LangFuse<br/>AI Observability"]
Traces["Agent Traces<br/>Cost Tracking"]
end
subgraph External["External Services"]
Claude["Claude<br/>Sonnet 4.5<br/>Haiku 4.5"]
OpenAI["OpenAI<br/>GPT-OSS-120B"]
Grok["Grok 2<br/>X AI"]
Gemini["Gemini 3 Pro<br/>Image Generation"]
Supabase["Supabase<br/>PostgreSQL<br/>Storage"]
end
React -->|HTTPS| FastAPI
FastAPI --> DocumentsAPI
FastAPI --> ChatAPI
FastAPI --> SearchAPI
FastAPI --> PropertyAPI
FastAPI --> FinancialAPI
FastAPI --> MaintenanceAPI
DocumentsAPI --> DocumentService
ChatAPI --> ChatService
SearchAPI --> SearchService
DocumentService --> TranslationService
DocumentService --> EmbeddingService
DocumentService --> QStash
QStash --> BackgroundTasks
EmbeddingService --> OpenAI
EmbeddingService --> PostgreSQL
BackgroundTasks --> DocumentsTable
ChatService --> Supervisor
Supervisor -->|Route| DocumentAgent
Supervisor -->|Route| FinancialAgent
Supervisor -->|Route| MaintenanceAgent
Supervisor -->|Route| PropertyAgent
DocumentAgent -->|RAG| SearchService
DocumentAgent -->|Query| DocumentsTable
FinancialAgent -->|Query| PostgreSQL
MaintenanceAgent -->|Query| PostgreSQL
PropertyAgent -->|Query| PostgreSQL
DocumentAgent --> Claude
FinancialAgent --> Claude
MaintenanceAgent --> Claude
PropertyAgent --> Claude
DocumentAgent --> Grok
FinancialAgent --> Gemini
Supervisor --> Checkpointer
Checkpointer --> CheckpointsTable
SearchService --> PostgreSQL
SearchService --> Redis
DocumentService --> LangFuse
Supervisor --> LangFuse
DocumentAgent --> LangFuse
FinancialAgent --> LangFuse
MaintenanceAgent --> LangFuse
PropertyAgent --> LangFuse
LangFuse --> Traces
style Frontend fill:#1e40af
style API fill:#059669
style Services fill:#7c3aed
style Agents fill:#dc2626
style SpecialistAgents fill:#ea580c
style Data fill:#6366f1
style Background fill:#0891b2
style Observability fill:#14b8a6
style External fill:#1C1C1E
Architecture: Clean Architecture (DDD) | LangGraph Multi-Agent System | Hybrid Search (Semantic + Full-Text) | QStash Background Jobs | LangFuse Observability | Redis Caching
🔒 Private projects - actively in development. Detailed architecture documentation available upon request.
I'm always open to discussing technology, architecture challenges, innovative projects, or collaboration opportunities.
"Code is like humor. When you have to explain it, it's bad." – Cory House