I help teams turn AI, data, and distributed-systems ideas into production systems people can actually use. My work usually sits where research prototypes, product requirements, and messy real-world infrastructure meet: agents connected to business APIs, retrieval systems grounded in private data, streaming and lakehouse platforms, and the evaluation loops that keep those systems reliable.

I have built systems across defense, cybersecurity, fintech, online marketplaces, secure communications, and autonomous navigation. I am comfortable moving between architecture, hands-on implementation, performance work, and the practical product decisions that make technical systems useful.

For consulting inquiries, email weiss.itamar@gmail.com.

What I help with

  • AI and machine learning in real-world applications and production systems
  • LLM agents that integrate with data, tools, APIs, and human workflows
  • Retrieval-augmented generation (RAG), embeddings, vector stores, and evaluation pipelines
  • Data lake, streaming, and analytics platforms using Spark, Kafka, Flink, Iceberg, Databricks, Snowflake, AWS, GCP, and Azure
  • Architecture reviews, technical due diligence, and implementation plans for complex engineering efforts
  • Product-facing prototypes that prove what is possible without ignoring production constraints

How I work

I am hands-on and code-first. I like to get close to the actual system, the data, the users, and the failure modes, then build the shortest path from a useful prototype to something robust enough for production.

I care about boring details because they are what make ambitious systems survive contact with reality: observability, evals, data quality, latency, costs, deployment ergonomics, and clear ownership boundaries. I also care about the product surface. A good AI or data system should not live only in a notebook, dashboard, or backend service; it should put capability directly in users' hands.

Selected work

AI, data, and infrastructure

Large-scale Apache Iceberg deployment. Deployed and optimized Apache Iceberg at a large-scale defense organization, enabling reliable, high-performance data lake management across petabytes of data. The project involved table-format architecture, schema evolution, and integration with existing data infrastructure.

Scalable Apache Flink pipeline for a cybersecurity startup. Built a real-time data pipeline and developed an internal framework on top of Flink, allowing teams to define complex, multi-stage streaming pipelines using a single YAML configuration file. The framework reduced development time and improved pipeline consistency and maintainability.

Database lineage graph optimization. Upgraded a recursive graph algorithm for computing database column-level lineage. The improved algorithm increased performance and accuracy, allowing the system to handle larger and more complex dependency graphs across enterprise-scale database environments.

Real-time Apache Kafka visualization. Created a tool to visualize real-time, high-throughput data streams, aggregate statistics, and provide real-time search capabilities using Node.js, Angular.js, Socket.io, and Apache Kafka.

Product and application systems

Online marketplace. Developed a two-sided marketplace where sellers could create profiles and offer services, while buyers could search, contact, book, pay online, and leave reviews. The system included admin workflows for approving sellers, monitoring transactions, handling disputes, and managing payouts.

SMS sending API. Developed a scalable SMS sending platform with a user-friendly API for a Kannel server. The platform included self-service subscription and payment flows, campaign management, and an admin interface for monitoring system health and activity.

Flash-card spaced repetition tool. Developed a Facebook app for creating and practicing flashcards using the spaced repetition algorithm.

Angular.js course. Designed and instructed a one-week Angular.js course, delivered multiple times to developers at Pictet Private Bank and other front-end developers in Geneva, Switzerland.

Security, blockchain, and autonomous systems

Accurate autonomous landing system for multicopters. Developed a computer-vision algorithm for identifying a unique pattern on a ground base station, enabling precise autonomous multicopter landing. The algorithm was designed to be accurate, computationally efficient, real-time, and height-invariant, and was implemented with OpenCV on an embedded Linux board.

Double-ratchet secure messaging system. Defined requirements for a secure messaging system, researched potential solutions, and selected, adapted, and integrated a double-ratchet messaging protocol into the client's application.

Blockchain customization. Customized a proof-of-stake blockchain for client-specific requirements, guiding multiple teams through the integration process. The resulting cryptocurrency passed several security and functionality audits.

Cryptocurrency exchange and Ethereum token. Developed a cryptocurrency exchange platform for trading between a custom cryptocurrency and Bitcoin, and created an Ethereum token with a smart contract for trustless exchange.

Background

I developed my first computer program in 1992 at the age of 10: an analog clock created using Logo. In high school I pursued a mix of creative and technical projects, including music production, 3D modeling, animation rendering, BBS management tools, cybersecurity experiments, radio-controlled model airplanes, a spaced repetition memorization tool in Prolog, and a natural language processing tool that could construct a semantic network and respond to natural-language queries.

I earned a B.Sc. in Physics and Mathematics from Hebrew University and an M.Sc. in Electrical Engineering from Tel Aviv University. That mix of mathematical modeling, engineering, and hands-on software work still shapes how I approach technical problems.

Education

  • B.Sc. in Physics and Mathematics (Talpiot) - Hebrew University (Dean's List 2002, 2003)
  • M.Sc. in Electrical Engineering - Tel Aviv University

I'm married to Dr. Sarah Weiss, neurologist specialized in stroke treatment.