Looking for a job

Shay Dan

@shaydan

Message

AI developer building Python CV + LLM pipelines, including agentic RAG and 3D volume estimation.

Israel

Message

What I'm looking for

I’d like to work on AI systems that combine computer vision with LLMs—especially agentic RAG and multi-agent orchestration—using well-tested Python pipelines and CI/CD from day one.

I build end-to-end Python pipelines that blend computer vision (detection, segmentation, instance segmentation) with LLM-based NLP, from 3D volume estimation to autonomous multi-agent systems.

At Infinity Labs R&D, I completed a 32-week, 3,000-hour intensive program—then shipped async pipelines with PyTest, Docker, and GitLab CI/CD, including a stateful LangGraph multi-agent system using MongoDB session persistence and asyncio concurrency.

I also developed an adaptive agentic RAG pipeline with LangChain, Gemini, cross-encoders, and Ragas for retrieval-quality evaluation, plus a YOLOv8 + SAM engine for fine-grained instance segmentation and 3D volume estimation.

Experience

Work history, roles, and key accomplishments

Current

Security Supervisor & Team Leader

Current

Modi'in Ezrachi

Jan 2016 - Present (10 years 7 months)

Led operational teams of 8+ personnel, handling strategic planning and security risk assessments on a part-time basis during studies.

AI Developer (Intensive Lab)

Infinity Labs R&D

Jan 2024 - Jan 2025 (1 year)

Completed a 32-week, 3,000-hour immersive program covering deep learning theory to end-to-end algorithmic implementation, building async Python pipelines with PyTest, Docker, and GitLab CI/CD.

Python PyTorch NumPy Pandas scikit learn SQL AsyncIO Pytest Docker Gitlab CI

Adaptive Agentic RAG Pipeline

Infinity Labs R&D

Built an async research-orchestration pipeline with autonomous query decomposition and an LLM-based gatekeeper for redundant-data filtering. Evaluated retrieval quality with Ragas, achieving 88% faithfulness.

Gemini Ragas LLM As Judge

Combat Engineering Officer (Lieutenant, Reserves)

IDF

Served as Platoon Commander leading 25+ soldiers in high-stakes field operations, managing logistics, operational risk, and real-time decisions under extreme pressure.

3D Object Volume & Caloric Estimation Engine

Infinity Labs R&D

Fine-tuned YOLOv8 and SAM for high-fidelity instance segmentation and 3D volume estimation from 2D images, achieving mAP50 of 0.91 across 20 classes with worst-case caloric error bounded to 26%. Applied PCA and clustering during EDA to identify data bias and drive targeted augmentations.

YOLOv8 SAM PyTorch OpenCV PCA Clustering EDA

Multi-Agent Scientific Health System

Infinity Labs R&D

Designed a stateful multi-agent system in LangGraph with MongoDB session-state persistence and asyncio for concurrent tool execution to cut pipeline latency. Integrated a custom RAG architecture over PubMed/arXiv corpora and built a Gradio interface.

LangGraph Gemini MongoDB AsyncIO Gradio RAG PubMed ArXiv