Superintelligence Lab - Members of Technical Staff

LG AI Research의 Superintelligence Lab - ICML candidate 직무 입사 지원서

채용 정보로 돌아가기 신규 Superintelligence Lab - ICML candidate Gangseo-gu, Seoul, South Korea

Member of Technical Staff

Agentic Systems, Multimodal Intelligence, and Efficient Inference

We are building the next generation of intelligent systems: agents that can reason, remember, search, use tools, interact with multimodal worlds, and execute complex work with reliability.

We are looking for Members of Technical Staff who can move between first-principles research and production-grade engineering. People who can attempt 0-to-1 ideas before they are obvious, but also care deeply about rigor, speed, quality, and systems that actually work.

This role is for builders who do not separate research from engineering. The work may become a paper, a product, an internal capability, an evaluation framework, or a new infrastructure layer. What matters is whether it advances the frontier.

• If you are already a senior staff, please contact moontae.lee@lgresearch.ai after submitting your application.

Areas of Focus

Agentic Orchestration

Design agents that can plan, delegate, reflect, verify, recover, and coordinate across tools, models, environments, and sub-agents.

Build architectures beyond simple chains: DAGs, recursive agents, hierarchical planners, multi-agent workflows, and adaptive execution systems.

Develop evaluation and verification methods for agentic behavior: trajectory analysis, tool-use evaluation, planning diagnostics, verifier models, reward signals, and failure recovery tests.

Create agents that are not only impressive in demos, but reliable under pressure.

Memory and Context Management

Build memory systems for agents that know what to keep, what to forget, what to retrieve, and what to compress.

Develop long-term memory, episodic memory, working memory, context compression, and retrieval-augmented reasoning systems.

Design memory evaluation methods that measure whether agents can reuse experience, maintain coherence, and improve over time.

Turn context from a limitation into a living substrate for intelligence.

Multimodal Intelligence

Develop and adapt VLMs, VLAs, image/video retrievers, and multimodal agent systems.

Build systems that can perceive, search, reason, and act over visual, embodied, and interactive environments.

Work with visual documents, diagrams, images/videos, tools, CAD/manufacturing, and real-world task trajectories.

Create multimodal evaluation methods for visual grounding, retrieval quality, embodied task execution, and complex multimodal work.

Systems and Efficient Inference

Build inference infrastructure for agentic workloads, not just single model calls.

Work on KV caching, speculative decoding, model routing, parallel decoding, long-context serving, retrieval latency optimization, and tool-call scheduling.

Design memory-efficient execution systems for agents that run across models, tools, retrievers, and environments.

Make frontier intelligence fast, scalable, inspectable, and reliable.

What You Will Do

Identify hard problems before they become standard benchmarks.

Prototype new ideas quickly, then turn the promising ones into serious systems.

Work across models, data, evaluation, infrastructure, and product-facing capabilities.

Build agents that can reason, search, remember, verify, and execute complex work.

Help define what a frontier agentic system should be.

You Might Be a Strong Fit If

You have built serious systems involving LLMs, agents, retrieval, multimodal models, inference infrastructure, or evaluation.

You are comfortable reading papers, writing code, running experiments, debugging systems, and questioning assumptions.

You can operate in ambiguity without waiting for a perfect problem statement.

You care about both conceptual novelty and engineering taste.

You have the courage to try strange ideas, and the discipline to kill weak ones.

You want your work to be measured not by activity, but by frontier movement.

Especially Relevant Backgrounds

LLM agents, planning, tool use, multi-agent systems, or agentic orchestration.

RAG, memory systems, context engineering, long-context modeling, or retrieval systems.

VLMs, VLAs, robotics, video understanding, multimodal retrieval, or embodied AI.

Model evaluation, verifier design, reward modeling, trajectory analysis, or benchmark construction.

Inference optimization, serving systems, distributed systems, compilers, databases, or search infrastructure.

Reinforcement learning, applied research engineering, or ambitious 0-to-1 system building.

What We Value

Taste over trend-following.

Depth over noise.

Speed with judgment.

Novelty with evidence.

Engineering with intellectual gravity.

High agency, high standards, and low ego.

The courage to build what does not yet have a name.

We are not looking for people who merely use frontier models. We are looking for people who can help build what comes after them.

채용 공고 알림 생성 LG AI Research 에서 경력을 쌓고 싶으신가요? 향후 채용 정보를 이메일로 바로 받아보세요. 알림 생성

이 직무 지원하기

*

필수 입력 필드를 나타냅니다

MyGreenhouse로 자동 완성

이름 *

성 *

이메일 *

전화 국가

전화

이력서/CV 파일 첨부 파일 첨부

Dropbox Google 드라이브 수동으로 입력 수동으로 입력

허용되는 파일 형식: pdf, doc, docx, txt, rtf

커버 레터 파일 첨부 파일 첨부

Dropbox Google 드라이브 수동으로 입력 수동으로 입력

허용되는 파일 형식: pdf, doc, docx, txt, rtf

LinkedIn Profile

Website

Optional Check 선택...

지원서 제출