Adi Yeltay

AI Research Engineering Intern

LibrAI

Built a macroeconomic briefing agent and evaluation loop, raising the composite quality score to 96/100; also developed a competition system that earned Gold and 2nd place.

• Built a Claude Agent SDK-based daily macroeconomic intelligence agent for Emirates Development Bank, delivering automated market briefings to analytics team • Integrated an AutoRubric LLM-as-a-judge evaluation framework into a fully automated, scheduled feedback loop, lifting composite quality score from 75.6 to 96/100 • Designed a multi-agent system with a meta-orchestrator and tailored retrieval knowledge base for the IOAI 2026 Model Track, securing a Gold Medal and 2nd place overall

Middle RPA Developer

Tele2

Telecommunications company.

Developed RPA agents and an on-premises trigger tool; automations saved more than 400 hours and $4,000 in monthly operating costs.

• Engineered 15+ RPA automation agents across departments, saving 400+ hours and $4,000+ in operating costs per month • Refactored legacy-system applications using Robotic Enterprise Framework (REFramework) • Built an end-to-end, on-premises, event-based C# tool that automatically triggered agent runs

AI/Software Engineer

PointAI

Built customer-support agents that resolved 80% of tickets without human intervention and migrated legacy code to Django REST Framework.

• Built custom customer-support agents in LangChain, ensembling GPT and Claude models to resolve 80% of tickets without human intervention • Migrated legacy code to Django REST Framework, improving reliability and scalability

Research Assistant

Internet of Things (IoT) Lab

Researched IoT communications, developed a protocol for six ESP32 devices, and published a paper at IEEE BlackSeaCom 2024.

• Published a paper in IEEE BlackSeaCom 2024 Conference • Developed a communication protocol for 6 ESP32 devices networked over a local Wi-Fi access point • Researched TS-LoRa communication technique to implement an algorithm to dynamically divide large file into packets based on each device’s unique parameters such as spreading factor, battery capacity, availability, and success rate

Master of Science, Natural Language Processing

Mohamed bin Zayed University of Artificial Intelligence

Bachelor of Science, Computer Science

Nazarbayev University