Agentic AI Intern

• Project support (Agentic Build): Assist with

gathering, prototype validation, and use case operations for agentic build initiatives.

• Data & experimentation: Curate datasets, maintain experiment logs, run evaluations, and summarize findings to inform iteration.

• Documentation & coordination: Prepare meeting notes and maintain use case repositories.

• Agent tools & multi model practice:Use mainstream agent/coding assistants (Claude Code, Codex) to prototype workflows and automate tasks; deeply use and compare LLMs (GPT, Gemini, Claude, Grok) in real tasks (work/learning/projects), iterate prompts/tool use strategies, and document best practices.

Candidate requirements

• Education: Bachelor’s or Master’s student; majors in Computer Science, Information Systems, Data Science, or Biomedical/Pharma-related fields are a plus.

• Core skills: Strong interest in Generative AI and agent frameworks; basic skills in Python/data handling or product operations are preferred; strong documentation and communication abilities.

• Evaluation mindset (Required): Ability to quantify value. Can define success metrics and run lightweight evaluations (e.g., success rate/accuracy, time saved, cost per task, human rating). Comfortable comparing baseline vs improved approach and summarizing conclusions.

• Agent tool practice (Preferred): Hands on practice using Claude Code or Codex in learning/projects (work experience is a plus). Able to clearly explain your workflow, what improved, and what you learned.

• LLM deep usage (Preferred): Deep hands on experience with at least 2 of GPT / Gemini / Claude / Grok, including prompt engineering, tool/function calling, structured outputs, long context handling, and quality iteration (evaluation or systematic feedback loops). Able to choose models based on task trade offs.

• Case study (Required): Provide at least 1 public or de identified case study (links/docs/demos are all acceptable; learning/projects are OK) that includes: problem → approach (agent tool + model(s) + prompt/tool strategy) → evaluation setup (data/criteria/baseline) → results (metrics) → takeaways.

• Work style: Detail oriented, accountable, and able to drive tasks to completion with clear status updates and retrospectives.

• Availability: On site at Shanghai Jing’an office at least 3 days per week; internship duration of at least 3 months (longer availability preferred).

• Nice to have: Experience with evaluation frameworks, multi agent workflows, dataset curation, or cross functional coordination in tech/product teams.