Чётко очерченная стажировка в сильной продуктовой компании даёт практический опыт в современных AI-системах, оценке агентов, безопасности и оптимизации моделей, но требует пяти дней в офисе.
Только офис
Актуальные AI-технологии
Настоящие задачи
Фокус на безопасности ИИ
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
AI Engineer, Intern (Python/PyTorch): Building and evaluating large-scale AI and agentic systems across data pipelines, model training, safety benchmarks, and production deployment with an accent on rigorous experimentation, benchmark design, and agent reliability. Focus on fine-tuning open-weight models, developing action-level safety evaluations, analyzing quantized models, and prototyping tool-calling agent systems.
Location: San Francisco, CA, US; in-office five days per week
Company
Postman is an API platform used by developers and organizations to build APIs and collaborate across the API lifecycle.
What you will do
Design benchmark tasks and mock API environments, extend Python evaluation harnesses and task-generation pipelines, and maintain a multi-model leaderboard.
Build an action-level AI safety benchmark covering prompt injection, data exfiltration, permission overreach, scenario design, threat modeling, and auditable evaluation.
Fine-tune open-weight models for tool calling and agentic tasks using SFT, distillation, and reinforcement learning with PyTorch and cloud GPU infrastructure.
Run benchmarked experiments, ablation studies, error analysis, experiment tracking, and cost-aware reporting.
Evaluate ultra-low-bit quantized models and analyze differences from full-precision models.
Prototype Postman’s in-product AI agent in TypeScript, including tool loops, multi-step execution, and checkpointing; document designs, experiments, runbooks, and safety concerns.
Requirements
Currently pursuing a BS, MS, or PhD in Computer Science, Data Science, or a related quantitative field.
Hands-on experience training or evaluating machine-learning models through coursework, research, hackathons, or an internship.
Solid Python fundamentals, including data structures, functions, basic testing, and writing or reviewing code outside notebooks.
Working knowledge of at least one deep-learning framework; PyTorch is preferred.
Clear written and verbal communication and a habit of documenting work.
Must work in the San Francisco office five days per week.
Nice to have
Experience fine-tuning open-weight LLMs with SFT, LoRA, reinforcement learning, or distillation and measuring improvements on benchmarks.
Experience building LLM agents or evaluation harnesses and reporting results with statistical rigor.
Experience shipping APIs, services, CLIs, Docker, CI/CD, or cloud software end-to-end; public GitHub code is a plus.
Interest or experience in AI safety, robustness, red-teaming, prompt injection, agent security, fairness, or interpretability.
Exposure to quantization, low-bit inference, or serving optimization.
Publications, technical blog posts, ablation studies, or self-directed projects with quantified results.
Fluency with AI coding tools such as Claude Code, Cursor, or Codex.
Culture & Benefits
Flexible schedule and collaborative, in-person work environment.
Full medical coverage, flexible PTO, wellness reimbursement, and a monthly lunch stipend.
Wellness programs, team-building events, and donation matching.
Inclusive culture emphasizing curiosity, transparency, honest communication, specific goals, and product quality.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →