AI Agent Engineer Intern: Build Self-Improving LLM Systems

TikTok; Singapore; On-site; SGD 17,000 - 21,000; Full time

Job summary

TikTok is seeking an AI Agent Algorithm Engineer Intern for Global E-Commerce, Conversational AI, starting 2027. You will build and evolve the agent runtime, orchestrating skills, tools, and memory across multi-turn conversations with guardrails.

Qualifications include pursuing an undergraduate or master's in CS/engineering, strong Python and one of C++, Go, Rust, or Java, plus solid ML/DL/NLP fundamentals and hands-on work with LLMs or agents.

Qualifications

• Currently pursuing an undergraduate or master's degree in software development, computer science, or related technical discipline.
• Strong Python plus one of C++ / Go / Rust / Java.
• Solid ML / DL / NLP fundamentals with hands-on experience with LLMs or agents.
• Able to read a paper or engineering blog and turn it into working code.

Responsibilities

• Build and evolve the agent runtime powering customer-service agents — orchestrate skills, tools, and context; implement loop control, intervention, and guardrails.
• Engineer context and memory for long multi-turn agents — memory, context summarization, and retrieval.
• Post-train and fine-tune LLMs (SFT / DPO / RL) and build data flywheel for training/eval/retrieval signals.
• Design and integrate tools, skills, and MCP connectors; enable tool search for large tool inventories.
• Build evaluation — LLM-as-judge with calibration; latency-aware harnesses; close offline-online gap.
• Build self-evolving loop — replay, user-simulator, Auto-RCA/Auto-GSB/Auto-Policy.
• Own an end-to-end surface and ship it to production across 30+ languages, measured on CSAT and containment rate.

Skills

Python

C++

ML/DL fundamentals

Education

Undergraduate or Master's degree in Computer Science or related

Job description

TikTok is seeking an AI Agent Algorithm Engineer Intern for Global E-Commerce, Conversational AI, starting 2027. You will build and evolve the agent runtime, orchestrating skills, tools, and memory across multi-turn conversations with guardrails.

Qualifications include pursuing an undergraduate or master's in CS/engineering, strong Python and one of C++, Go, Rust, or Java, plus solid ML/DL/NLP fundamentals and hands-on work with LLMs or agents.