Mercor
Mercor
51 – 200 Darbinieki
Mercor pieejamajā profilā nav pietiekami daudz pārbaudītas informācijas par uzņēmuma produktiem, pakalpojumiem, klientiem, darbības nozari vai personāla atlases fokusu, lai sagatavotu precīzu uzņēmuma aprakstu. Lai aprakstītu uzņēmumu vai tā piedāvātās iespējas, nepieciešams papildu avota materiāls.

Remote Machine Learning Research Expert

Develop challenging inference, evaluation, and systems task ideas for Mercor’s frontier AI research projects. Focus on problems that push advanced models to their limits.

Apraksts

  • Propose difficult machine learning engineering tasks for a long-term research project with a leading AI lab.
  • Create challenges that frontier models solve in no more than two of eight attempts.
  • Concentrate on inference, evaluation, and systems rather than extensive model training.
  • Build a roughly even mix of GPU and CPU-only task ideas.
  • Explain clearly what makes each problem difficult for a frontier model.
  • Contribute mainly ideas; task implementation and review are not required.

Prasības

  • Currently or recently worked as an ML research engineer, research scientist, or applied scientist.
  • Hands-on experience personally training, fine-tuning, or optimizing models with PyTorch or JAX.
  • Deep expertise in at least one area: LLM pretraining, alignment, or evaluation; GPU inference and performance, including vLLM, Triton, CUDA, or profiling; reinforcement learning; or scientific ML in weather and climate, physics, or biology.
  • Can explain why a problem is difficult for a frontier model.
  • Benchmark or evaluation question writing experience is a plus.
  • A PhD in ML, computer science, or physics, or an MSc paired with strong research output, is a plus.
  • Publications are a plus.
  • Strong performance on previous Mercor contracts is a plus.
  • H1-B and STEM OPT candidates are ineligible because Mercor cannot support them.
  • Software or full-stack engineers whose AI experience is mainly LangChain or RAG applications are not a fit.
  • Candidates whose experience is limited to AI trainer, rater, or annotator work are not a fit.

Priekšrocības

  • Fully remote work.
  • Flexible schedule.
  • Weekly payments via Stripe or Wise for services rendered.
  • Projects may be extended, shortened, or ended early based on needs and performance.
  • Competitive pay.
  • Work with leading researchers.
  • Help shape the next generation of AI systems.
  • Earn up to $800 per successful referral, subject to referral limits.
  • Reasonable accommodations are available upon request.

Saistītās vakances

Gramian Consulting

Software Engineer for AI Code Licensing Program

Gramian Consulting

Submit an existing software repository for assessment of its licensing value. Eligible code owners may receive compensation under agreed non-exclusive licensing terms.

Atvērt