Data Science AI Engineer

Company Description

Clears AI is an autonomous software execution platform designed to transform AI-accelerated coding into real and measurable delivery outcomes. By automating workflows across the software development lifecycle (SDLC), Clears replaces manual coordination with workflows that enhance requirements, synchronize efforts between AI and humans, and provide ongoing insights into execution and processes at scale. This system-focused approach helps engineering teams transition from using isolated AI coding tools to achieving impactful, organization-wide productivity. Clears is dedicated to driving innovation and efficiency through the adoption of AI in software development.

Role Description

This is a full-time, hybrid role for a Data Science AI Engineer based in Herzliya. The Data Science AI Engineer will design, build, and evaluate the AI systems that power Clears’ autonomous software execution platform. This includes developing agentic retrieval systems, improving planning and reasoning workflows, building evaluation frameworks, and optimizing LLM-based execution for production reliability.

You will work on complex AI problems at the intersection of data science, retrieval, agents, and software engineering. The role requires strong hands-on ability with Python, deep understanding of LLM and agent behavior, and the ability to turn experimental AI workflows into measurable, production-grade systems.

Qualifications

• 7+ Years Experience: Deep professional background in AI, data science, or applied machine learning.
• LLM & Agent Behavior: Deep understanding of how LLMs and agents behave in real-world workflows, including reasoning patterns, hallucination risks, context degradation, tool-calling failures, and methods for improving consistency and reliability.
• Agent Expert: Proven experience building high-quality agentic systems, including planning, tool-calling, self-correction loops, and multi-step reasoning workflows.
• Sophisticated Retrieval: Strong hands-on experience with advanced retrieval systems, including vector indexes, hybrid search, GraphRAG, codebase retrieval, and agentic retrieval patterns.
• Evaluation-Driven: Experience building LLM evaluation frameworks and measuring quality using precision, recall, success rate, failure analysis, regression testing, and other structured eval practices.
• Prompt & Performance Optimization: Strong ability to improve LLM behavior using prompt engineering, few-shot learning, structured outputs, and targeted fine-tuning where relevant.
• Technical Excellence: Expert-level Python. Able to bridge the gap between model behavior, data pipelines, retrieval systems, and production backend services.
• AI Native: Power user of Claude, Cursor, Claude Code, and modern AI development workflows.
• Production Mindset: Experience moving agentic systems from impressive demos to reliable, measurable, production-grade systems.