TOMER BITON

2026
LinkedIn LinkedIn Email Email

Selected Projects

Portfolio  /  Vol. I

Building intelligent data systems and scalable multi-tenant platforms, integrating LLM-powered analytics with structured and unstructured data sources. Specialized in API architecture, secure access control (RBAC/RLS), and production-grade backend system design.

01 - Primary Project
Intelligence  /  AI  /  Data Fusion

OSINT Fusion Data Pipeline

Satellite + News + AI Forecasting Architecture

Designed and implemented a multi-layer data fusion system integrating satellite imagery (GEOINT), multi-source news ingestion (OSINT), transformer-based classification, and probabilistic confidence modeling into structured intelligence outputs.

Architecture Flow

Request Data Collection Normalization NLP Classification Signal Aggregation Confidence Scoring REST API Output

Modular pipeline orchestration - collectors, forecasting engine, confidence module & controller

Multi-source RSS ingestion & structured NewsArticle normalization

Transformer-based zero-shot event classification

Weighted confidence fusion engine - source count, contradiction score, classification probability, signal strength

Risk-profile adaptive forecast generation

FastAPI REST backend with high-throughput async processing

Satellite imagery integration (GEOINT) with spatial indexing

Background job execution with persistent status tracking

Structured report generation - JSON + Markdown dual output

PythonFastAPITransformers REST APIsData ModelingPipeline ArchitectureSignal Aggregation
02
Analytics  /  LLM  /  Multi-Tenant

AI-Powered KPI & Data Insights Platform

Multi-Database Analytics & LLM-to-SQL System

Built a multi-tenant analytics platform integrating structured and unstructured data sources, enabling natural language querying via LLM-based SQL generation and delivering real-time KPI dashboards with automation triggers.

LLM → SQL transformation layer with dynamic schema mapping & field normalization

Multi-database connector registry - Postgres, Supabase, MongoDB

Row-level security enforcement & granular RBAC implementation

Webhook-based automation triggers for reactive workflows

Multi-tenant architecture with full data isolation

Interactive KPI dashboard with dynamic filtering

Structured data abstraction layer for cross-database querying

Modular REST API backend with async processing

PythonFastAPIPostgreSQL SupabaseMongoDBLLM IntegrationMulti-Tenant Systems
03
Fullstack  /  Workflow

Enterprise Knowledge
Retention System

Relational Data Modeling & Access-Controlled Workflow Platform

ERD-driven schema design - MySQL relational modeling

Multi-layer RBAC - client + server validation enforcement

Two-way GraphQL integration with external PM system

OAuth calendar scheduling & version control workflow

Kanban state transitions, CSV export, Chart.js dashboards

Session-based auth & environment configuration management

Node.jsExpressMySQL GraphQLOAuthRBAC
04
OCR  /  Document Processing

Invoice OCR & Structured
Extraction Pipeline

Unstructured Document → Validated Structured JSON

OCR integration - image & PDF invoice ingestion via EasyOCR

Text normalization & numeric correction logic

Regex-based field extraction - dates, totals, tax, line items

Dynamic line-item reconstruction with table parsing

JSON serialization with schema validation layer

PDF summary generation from structured output

PythonEasyOCRRegex Data CleaningJSON SerializationDocument Processing
05
NLP  /  Computer Vision  /  Transformers

Applied Transformer Model Implementations

NLP & Vision inference pipelines - model loading, batch processing, image preprocessing, structured output formatting

DistilBERT
Sentiment classification & NLU inference
Vision Transformer (ViT)
Image classification & embedding pipelines
CLIP
Multimodal vision-language alignment
Donut
Document transformer - structured document understanding
PythonHuggingFace Transformers PyTorchNLPComputer Vision
Core Skills Demonstrated Data Engineering & System Design
Data Pipeline Architecture
Multi-Source Aggregation
API Design & Integration
Database Modeling
LLM Integration
Transformer-Based NLP
Signal Fusion & Confidence
Role-Based Access Control
Workflow State Management
Multi-Tenant Architecture
REST & Async Processing
Unstructured Data Processing