Likeremote

Subscribe to the latest remote jobs:

  • Likeremote jobs on https://LinkedIn.com/
  • Likeremote jobs on https://telegram.org/
  • Likeremote jobs on Reddit.com
4

Senior AI Engineer (Driving VLM/VLA)

42dot
πŸ‡°πŸ‡· South Korea
Hybrid
Senior
1 month ago
  • AI
  • Temporal
  • Computer Vision
  • Machine Learning
  • PyTorch
Not scoredNo CV on file. Upload one and this job gets a score out of 100.Upload CV

About the Team & Mission

42dot의 Senior AI Engineer (Driving VLM/VLA)λŠ” 수백만 건의 μ‹€μ œ μžμœ¨μ£Όν–‰ 데이터λ₯Ό ν™œμš©ν•΄ μ°¨μ„ΈλŒ€ Driving Foundation Model을 κ°œλ°œν•©λ‹ˆλ‹€. Vision-Language-Model(VLM), Vision-Language-Action(VLA), λ©€ν‹°λͺ¨λ‹¬ AI κΈ°μˆ μ„ 톡해 μ°¨λŸ‰μ΄ λ³΅μž‘ν•œ λ„λ‘œ ν™˜κ²½μ„ μ΄ν•΄ν•˜κ³  νŒλ‹¨ν•˜λ©° 행동할 수 μžˆλ„λ‘ λ§Œλ“œλŠ” 핡심 역할을 μˆ˜ν–‰ν•©λ‹ˆλ‹€. λŒ€κ·œλͺ¨ GPU 인프라와 λ…μžμ μΈ μžμœ¨μ£Όν–‰ 데이터셋을 기반으둜 driving scene understanding, multimodal reasoning, VLM/VLAλ₯Ό 톡해 μžμœ¨μ£Όν–‰ μ‹œμŠ€ν…œμ˜ 인지, νŒλ‹¨, κ³„νš μ„±λŠ₯을 κ³ λ„ν™”ν•©λ‹ˆλ‹€.

Responsibilities

  • μžμœ¨μ£Όν–‰ 데이터λ₯Ό ν™œμš©ν•œ Vision-Language-Action 기반 driving foundation model 섀계 및 개발

  • Camera/video, map, trajectory, action, language λ“± multimodal sequential data 기반 λͺ¨λΈ ν•™μŠ΅ 및 평가

  • Driving scene understanding, temporal reasoning, agent interaction modeling, risk/event understanding λͺ¨λΈ 개발

  • Scene captioning, visual question answering, auto-labeling, data mining, retrieval λ“± VLM 기반 μ‘μš© κΈ°λŠ₯ 개발

  • λŒ€κ·œλͺ¨ multimodal dataset ꡬ좕, ν•™μŠ΅ recipe, evaluation benchmark, ablation μ‹€ν—˜ 섀계

  • Simulation, Data, ML Platform 쑰직과 ν˜‘μ—…ν•˜μ—¬ λͺ¨λΈμ„ μžμœ¨μ£Όν–‰ μ‹œμŠ€ν…œμ— 톡합

Qualifications

  • Computer Vision, Machine Learning, Robotics, Autonomous Driving, Multimodal AI κ΄€λ ¨ 5λ…„ μ΄μƒμ˜ 연ꡬ/개발 κ²½ν—˜ λ˜λŠ” 이에 μ€€ν•˜λŠ” μ—­λŸ‰ λ˜λŠ” Foundation Model, Multimodal Learning, Generative AI λΆ„μ•Όμ—μ„œμ˜ λ›°μ–΄λ‚œ 연ꡬ μ„±κ³Ό(proven track record)λ₯Ό λ³΄μœ ν•˜μ‹  λΆ„

  • PyTorch 기반 deep learning model 개발 및 ν•™μŠ΅ κ²½ν—˜

  • Vision model, video model, VLM, VLA, world model, trajectory prediction, imitation learning 쀑 ν•˜λ‚˜ 이상에 λŒ€ν•œ κΉŠμ€ 이해와 κ΅¬ν˜„ κ²½ν—˜

  • 이미지/λΉ„λ””μ˜€/sensor/trajectory/language/action λ“± multimodal λ˜λŠ” sequential data 처리 κ²½ν—˜

  • Transformer, diffusion, autoregressive model, representation learning 쀑 ν•˜λ‚˜ 이상에 λŒ€ν•œ 이해

  • λ…Όλ¬Έ 기반 아이디어λ₯Ό κ΅¬ν˜„ν•˜κ³  λŒ€κ·œλͺ¨ λ°μ΄ν„°μ—μ„œ ν•™μŠ΅, 평가, κ°œμ„ ν•œ κ²½ν—˜

  • 문제λ₯Ό λ…λ¦½μ μœΌλ‘œ μ •μ˜ν•˜κ³ , μ‹€ν—˜ 섀계뢀터 λͺ¨λΈ κ°œμ„ κΉŒμ§€ μ£Όλ„μ μœΌλ‘œ μˆ˜ν–‰ν•  수 μžˆλŠ” μ—­λŸ‰

Preferred Qualifications

  • μžμœ¨μ£Όν–‰ λ˜λŠ” robotics foundation model 개발 κ²½ν—˜

  • VLM/LLM fine-tuning, instruction tuning, multimodal alignment κ²½ν—˜

  • VLA, behavior cloning, imitation learning, trajectory planning, policy learning κ²½ν—˜

  • Action-conditioned video generation, world model, 4D scene modeling κ²½ν—˜

  • BEV, occupancy, map, trajectory, agent interaction λ“± driving-specific representation κ²½ν—˜

  • Distributed training, large-scale multimodal data pipeline, data mining pipeline κ²½ν—˜

  • Closed-loop simulation λ˜λŠ” scenario-based evaluationκ³Ό λͺ¨λΈμ„ μ—°κ²°ν•΄λ³Έ κ²½ν—˜

Interview Process

  1. μ„œλ₯˜ μ „ν˜•

  2. μ½”λ”© ν…ŒμŠ€νŠΈ

  3. 1μ°¨ λ©΄μ ‘ (화상, 1μ‹œκ°„ λ‚΄μ™Έ)

  4. 2μ°¨ λ©΄μ ‘ (λŒ€λ©΄ ν˜Ήμ€ 화상, 3μ‹œκ°„ λ‚΄μ™Έ)

  5. 처우 ν˜‘μ˜Β·μž…μ‚¬

Additional Information

  • μ „ν˜• μ ˆμ°¨λŠ” 일정 및 μ§„ν–‰ 상황에 따라 일뢀 변경될 수 있으며, 각 μ „ν˜• κ²°κ³ΌλŠ” λ“±λ‘ν•˜μ‹  μ΄λ©”μΌλ‘œ κ°œλ³„ μ•ˆλ‚΄λ“œλ¦½λ‹ˆλ‹€.

  • μ§€μ›μ„œ 제좜 μ‹œ μ£Όλ―Όλ“±λ‘λ²ˆν˜Έ, 가쑱관계, 혼인 μ—¬λΆ€, 연봉, 사진, 신체쑰건, μΆœμ‹  μ§€μ—­ λ“± μ±„μš©μ ˆμ°¨λ²•μƒ μš”κ΅¬ κΈˆμ§€λœ μ •λ³΄λŠ” μ œμ™Έ λΆ€νƒλ“œλ¦½λ‹ˆλ‹€.

  • μ§€μ›μ„œ μ ‘μˆ˜ 쀑 였λ₯˜κ°€ λ°œμƒν•˜κ±°λ‚˜ 기타 문의 사항이 μžˆμ„ 경우, recruit@42dot.ai둜 λ¬Έμ˜ν•΄ μ£Όμ‹œκΈ° λ°”λžλ‹ˆλ‹€.

  • κ΅­κ°€λ³΄ν›ˆλŒ€μƒμž 및 μ·¨μ—…λ³΄ν˜Έ λŒ€μƒμžλŠ” 관계법령에 따라 μš°λŒ€ν•©λ‹ˆλ‹€.

  • μž₯애인 고용 촉진 및 μ§μ—…μž¬ν™œλ²•μ— 따라 μž₯애인 등둝증 μ†Œμ§€μžλ₯Ό μš°λŒ€ν•©λ‹ˆλ‹€.

  • 42dot은 μ˜λ’°ν•˜μ§€ μ•Šμ€ μ„œμΉ˜νŽŒμ˜ 이λ ₯μ„œλ₯Ό λ°›μ§€ μ•ŠμœΌλ©°, μš”μ²­ν•˜μ§€ μ•Šμ€ 이λ ₯μ„œμ— λŒ€ν•΄ 수수료λ₯Ό μ§€λΆˆν•˜μ§€ μ•ŠμŠ΅λ‹ˆλ‹€.

  • μ§€μ›μ„œ λ‚΄μš© 쀑 ν—ˆμœ„ 사싀이 발견될 경우, μž…μ‚¬κ°€ μ·¨μ†Œλ  수 μžˆμŠ΅λ‹ˆλ‹€.

  • 인터뷰 ν”„λ‘œμ„ΈμŠ€ μ’…λ£Œ ν›„ μ§€μ›μžμ˜ λ™μ˜ν•˜μ— ν‰νŒμ‘°νšŒκ°€ 진행될 수 μžˆμŠ΅λ‹ˆλ‹€.

  • 3κ°œμ›”μ˜ μˆ˜μŠ΅κΈ°κ°„μ΄ 적용될 수 μžˆμŠ΅λ‹ˆλ‹€.

Senior AI Engineer (Driving VLM/VLA) Β· 42dot

Auto apply with Likeremote