Senior Architect, Model Performance and Post-Training

This is a fully remote position, open to applicants in California, +1 more state.

📋 Description

• Create detailed frameworks to assess and scrutinize the performance of coding models related to code generation, debugging, code completion, repository-level reasoning, and multi-step software engineering tasks.

• Design and enhance post-training workflows for coding models, which encompass supervised fine-tuning (SFT), preference optimization, reinforcement learning (RL), as well as outcome-based or process-based reward modeling.

• Evaluate benchmark results, training dynamics, failure modes, and model behaviors to identify areas for enhancement.

• Elevate model performance on complex coding tasks through focused training interventions, reward structuring, curriculum development, and iterative error analysis.

• Develop scalable and reproducible infrastructure for conducting training experiments, evaluation pipelines, and model performance assessments.

• Work collaboratively with research scientists, infrastructure engineers, and product teams to translate model insights into production-ready functionalities.

⛳️ Requirements

• Bachelor's degree (or equivalent experience) in Computer Science or AI.

• Over 12 years of relevant work experience.

• Proficient programming skills in Python.

• Strong grasp of large language models, transformer architectures, and the training and evaluation processes for generative models.

• Fundamental understanding of reinforcement learning algorithms and their implementations, including MDPs, policies, and reward design.

• In-depth knowledge of software engineering principles, code generation, automated testing, and program correctness.

• Proven ability to oversee model development life cycles, including dataset versioning, experiment tracking, and evaluation pipelines.

• Experience in training or post-training state-of-the-art coding models or models designed for software engineering agents.

• Familiarity with execution-based rewards, verifiable rewards, reward hacking mitigation, and automated code assessment.

• Knowledge of coding benchmarks and evaluation environments like SWE-bench.

🏝️ Benefits

• Highly competitive salaries.

• Comprehensive benefits package.

• Equity options.

• Benefits for you and your family.