AI, Automation & Software · Business Systems
Hi, I'm
I build AI agents and automations that take real work off people's plates and turn scattered business data into decisions leaders actually act on.
The report someone rebuilds by hand every Monday. The leads that sit unread in an inbox. The data spread across five systems that never quite agree.
I'm an AI and automation engineer based in Israel. I build the agents, pipelines and integrations that make those problems quietly disappear, and I care as much about the business outcome as the code that gets there.
Agents, RAG and LLM workflows with Claude and OpenAI that read, decide and act, so people don't have to.
I connect CRMs, ERPs and data sources, then turn the numbers into something leadership can actually steer by.
Full-stack and cloud, from a blank repo to a live URL. If it needs to ship, I ship it.
A mix of production systems, client work and things I made because it's cool when you can scratch your own itch.
React and TypeScript up front, AWS Lambda, DynamoDB and WebSockets underneath, on a live custom domain. Built to scale, not just to demo.
Claude verification layered on deterministic glossary checks and quality scoring, with anything questionable routed automatically into Asana for human review.
An n8n integration syncing e-commerce data into a CRM in real time, deployed on always-on cloud infrastructure.
Claude and Python pipelines that pull, structure and summarise large datasets into forecasting and market-intelligence views in Power BI, feeding real GTM and strategy calls.
An automation that enriches inbound conference and web leads and routes them to the correct owner, cutting the lag between interest and follow-up.
A zero-knowledge sealed-bid auction built with a partner in Rust on the RISC Zero zkVM, proving the winning bid is correct without revealing anyone's bid. A serious dive into applied cryptography.
A personal build that turns workouts into cards you drag across columns — unstarted, in progress, done — so a training week reads like a kanban board. It layers in gamification too: micro-rewards and streaks for finishing sets, to make showing up feel good. Equal parts side project and my own daily driver in the gym.
Built with Google Apps Script: real-time movement, score tracking and a top-score leaderboard, all rendered right in the cells of a Google Sheet.
A year of production automations at Grapa gave back roughly 975 hours — nearly half a full-time role. Here's where the time went, and what it's worth.
Nearly half a full-time role in recovered capacity, against a 9-hour day, 5-day week. Net of build & maintenance, it lands near 825 hours.
A workflow only saves time if people actually adopt it, and that part is human, not technical.
The hours above are only real because my colleagues use these tools every day. Getting there meant sitting with people, understanding how they actually work, and walking them through each automation patiently, at their pace and not mine, until it fit naturally into their day instead of feeling like one more thing to learn.
It's also why I teach. I run AI-implementation sessions one on one and in groups, and record the walkthroughs as slide decks and Loom videos, so anyone can come back and re-watch exactly how a workflow runs whenever they need a refresher.
Watching someone stop dreading a task they used to do by hand, and get that time back for work that matters more, is the part I find most rewarding. Building the automation is the easy half; helping people trust it and work more efficiently with it is what makes the impact last.
Conservative per-task estimates: hours = (manual minutes × runs per year) ÷ 60, rounded down. Money is illustrative — 975 hours at a ~₪120/hour loaded cost; adjust to your rate.
Building internal AI agents and automation across CRM, ERP and analytics, and working directly with the CEO on research, market intelligence and forecasting that feeds executive strategy.
Automated internal workflows and built custom tools with Python, Apps Script and Salesforce, integrating NetSuite, Google Workspace and Mailchimp across international teams.
Kept production SQL data clean and reliable, wrote scripts for validation, migration and cleanup, and worked directly with customers to resolve live data issues.
Built Python data pipelines turning raw banking data into model-ready inputs for AI-driven risk and prediction models.
Bringing AI into a business is only worth it if it never becomes a liability. At Grapa, making sure it doesn't is part of my job.
I check what every AI tool sends, stores and trains on, and keep sensitive company and customer data out of places it shouldn't go.
Human review on anything high-stakes, least-privilege access, and clear limits on what agents are allowed to see and do.
I assess each vendor's security posture and compliance before anything is wired into our systems, and advise leadership on the trade-offs.
I train often and take my health seriously. Black belt in judo, strength work, and a real interest in the science of how the body moves, adapts and grows. Physical health supports the mind, and it's a big part of what keeps me sharp.
Most of my projects start one of two ways: "it would be cool to…" or my wife asking "can you build me this?" Snatch Words began exactly that way. I just like making things, so plenty of them happen on weekends.
My wife and daughter are what it's all for. They're the reason I stay disciplined and the calm when work gets loud. And my dog's pretty cute too.
Looking for someone who can take an AI or automation idea from "wouldn't it be great if" all the way to production? That's the part I like most.
