Israel's job board Browse all jobs

Research Engineer

Qodo · Telaviv

Description Foundation Group takes on the open research problems behind Qodo's agentic stack and turns them into solutions that make the whole stack stronger, from the code review agents customers use today to the systems we ship next. We are looking for a research engineer to own problems in that space end to end, from the research question to production: frame the question, build the data and the experiments, prove the gain on benchmarks, and ship the result to customers. The role spans sleep-time compute, multi-agent harnesses, self-learning agents, context engineering, adaptive classification, model training and more. You will take on problems across these areas as the stack grows, and take a model or an agent and make it measurably better. Responsibilities Weigh research bets against product impact, set the approach and milestones, and stay accountable for the outcome once it runs in production. Identify the signals that show whether an agent is doing its job in production, decide what context it needs and how to supply it, and run A/B tests that prove a change gives a statistically significant gain. Build the feedback loops that let Qodo's core products learn from their own use: find ways for agents to improve themselves from outcomes and traces, and make every step measurable so the products get better in a continuous improvement cycle. Curate datasets and build offline benchmarks and evaluation suites for code review agents that the team trusts, with ablations and reproducible results, and use them to decide what ships. We designed and published the Qodo Code Review Benchmark, and extending it is part of this job. Stay current with the latest research and agentic patterns, test what is new against our problems, and bring in what works so Qodo's stack never falls behind. Requirements 5+ years of hands-on ML experience, at least 2 of them with LLMs, on projects you owned from problem definition to production use. A strong NLP and generative AI background, with hands-on experience training or fine-tuning language models (supervised fine-tuning, reinforcement learning or reinforcement fine-tuning, distillation), not only prompting them. Experience building or evaluating LLM agents: tool use, planning, retrieval, multi-agent orchestration. Experience building datasets and evaluation benchmarks for LLM applications or agents. Strong software engineering skills to take research initiatives to production. MSc in Computer Science, Mathematics or a related field. A PhD is a plus. Clear written and spoken communication of results, including negative ones. Nice to have A reinforcement learning background: reward design, offline and online methods. Publications, open-source contributions or public benchmarks. Experience with managed training platforms, or with running training on your own GPUs.

Apply »

More jobs at Qodo

All jobs at Qodo »

Similar jobs

Find your next role on Israel's job board.

Browse all jobs