Abdelrahman Yosef

Conducted 5,000+ RLHF evaluations (93% acceptance rate) to fine-tune major frontier models on complex code generation and logic workloads. Audited end-to-end agentic workflows and function-calling capabilities, systematically identifying hallucinations and API failures.

RLHF Code Expert at Outlier (2024-02 – Present)

• Conducted 5,000+ RLHF evaluations (93% acceptance rate) to fine-tune major frontier models on complex code generation and logic workloads.
• Audited end-to-end agentic workflows and function-calling capabilities, systematically identifying hallucinations and API failures.

Eval & Benchmarking ML Researcher at Alignerr (2025-01 – Present)

• Designed adversarial ML research tasks with deterministic autograders, providing high-signal RL feedback loops to evaluate state-of-the-art LLMs.
• Built multimodal evaluation pipelines across image and audio domains, creating robust detection systems for cross-modality reasoning failures.

MERN Developer Lead at University of Rochester, Information Technology (2019-08 – 2023-12)

• Led MERN platform development on AWS for 12K+ users, reducing MTTR to <5m via zero-downtime Docker deployments.
• Improved p95 load speeds from >3s to <300ms via server-side pagination, query optimization, and in-memory caching.
• Integrated Shibboleth SSO, dropping support tickets by 80%, and rightsized AWS resources to cut costs by 25%.

M.S. in AI for Business – University of Rochester, Simon Business School (2026-05)

B.S. in Computer Science, Minor in Business – University of Rochester (2024-05)