available for opportunities ↗
building
ai systems
& real products
Third-year Information Systems student at Yezreel Valley College. Specializing in AI agent architectures and full-stack development.
About me
Third-year Information Systems student at Yezreel Valley College, Israel, specializing in AI agent architectures and full-stack development.
I design tool-based AI systems including MCP server architectures and LLM-powered agents and build production-grade applications with React, TypeScript, and Python. My foundation includes strong OOP, data structures, QA methodology, and system design. Currently building toward my first role as an Applied AI Engineer.
Education
B.Sc. Information Systems
Yezreel Valley College, Israel · 3rd Year
Based in
Israel 🇮🇱
Impressive Works
A selection of projects that showcase my work in AI and engineering.
Gamified task management system that turns your to-do list into an RPG quest board.
- ·Turns tasks into RPG quests — earn XP, levels, and rewards for real work
- ·Global state via Zustand — lightweight and scalable without Redux overhead
- ·Jira API integration planned to sync real work tasks into the quest board
Tool-based AI agent for transaction categorization, monthly reports, and budget suggestions using MCP architecture.
- ·MCP tool-based agent with 5 discrete tools — the LLM orchestrates, not guesses
- ·Clean tool separation makes each function independently testable
- ·Mirrors how production AI systems are actually architected
LLM-powered summarization tool built on an agentic architecture for intelligent document processing.
- ·Agentic summarization — the LLM decides how to chunk and prioritize, not just truncate
- ·Handles variable-length documents with structured, consistent output
- ·Document-type awareness formats output based on content structure
Artist marketplace with buyer profiles, social feed, and a virtual gallery concept for discovering independent creators.
- ·Combines social feed with artist profiles — discovery meets self-expression
- ·Gallery-first UI designed for independent artists and art buyers
- ·Future concept: 360° virtual gallery rooms per artist
Skills that fuel
my passion
Languages
Writing clean, efficient code across multiple paradigms.
Frontend
Building responsive, interactive UIs with modern frameworks.
Backend / DB
Designing scalable APIs and data persistence layers.
AI / ML
Architecting agent-based systems and LLM-powered tools.
DevOps / Tools
Deploying and managing cloud-native infrastructure.
QA
Ensuring software quality through rigorous testing practices.
AI Tooling
Beyond using AI
building the tools
that make AI more capable.
I design and build SKILL.md files structured instruction sets that give Claude Code a specialized mode for a given domain. Each skill encodes a workflow, quality bar, and output standard.
/agent-planner
AI Engineering
Designs agents and multi-agent systems. A 5-phase workflow problem framing, tool design, orchestration pattern, failure handling, output artifact that guides from a vague idea to a concrete architecture. Distinguishes when to use an agent vs. a simpler prompt chain, and surfaces latency/cost tradeoffs at every step.
/ai-cost-optimizer
AI Engineering
Audits LLM spending and reduces API costs with real dollar numbers. Covers model right-sizing, prompt caching, Batch API, output length control, and semantic caching. Works from a baseline cost calculation and ranks optimizations by ROI never gives a recommendation without attaching a number.
/skill-forge
AI Engineering
Builds or audits Claude Code skills. Two modes: BUILD (new skill from scratch with qualification gates, interview, design proposal, and CLAUDE.md integration) and AUDIT (scores an existing skill across 9 dimensions against a quality rubric, then shows before/after diffs for every gap).
/explain-code
Developer Experience
Explains any codebase using analogies, ASCII flow diagrams, step-by-step walkthroughs, and a common gotcha. Makes complex code understandable at the right level not just what the code does, but why it's built that way.
This reflects a tooling-level understanding of AI systems not just prompt-level interaction.
That's all for now.