Research Engineer, Large Language Models

4 hours, 42 minutes ago
Full-time
Mid Level
Software Development
Immiland Canada

Immiland Canada

Immiland Canada is a comprehensive immigration consulting service that provides advice and assistance for individuals and families looking to immigrate to Canada. They offer services for work permits, study permits, education agreements, and support fo...

Family Services
11-50
Founded 2016

Description

  • Design and run large-scale pre-training and fine-tuning experiments, from ablations to full training runs.
  • Study scaling laws to select model size, data mix, and hyperparameters within compute budgets.
  • Improve training efficiency using distributed parallelism, mixed precision, and custom CUDA or Triton kernels.
  • Diagnose and resolve training instability, divergence, loss spikes, and hardware failures.
  • Build and improve data pipelines for filtering, deduplication, quality scoring, and dataset mixing.
  • Implement and validate ideas from recent research papers at scale.
  • Build capability evaluations and rigorously track experimental results.
  • Collaborate with alignment, infrastructure, and product teams and communicate findings through written reports.

Requirements

  • 3+ years of hands-on deep learning model-training experience, or a PhD in computer science, mathematics, physics, or a related field.
  • Strong Python skills and deep expertise in PyTorch or JAX.
  • Practical experience with Transformer architectures, including attention, tokenization, embeddings, and positional encodings.
  • Experience with distributed training on multi-node GPU clusters using FSDP, DeepSpeed, Megatron-LM, or similar tools.
  • Strong foundations in linear algebra, probability, statistics, and optimization.
  • Ability to implement research papers from scratch and evaluate their effectiveness.
  • Rigorous experimentation practices, including baselines, ablations, reproducibility, and experiment tracking.
  • Clear written and verbal communication skills and comfort working in an ambiguous, fast-moving environment.
  • Publications at NeurIPS, ICML, ICLR, or ACL, or notable open-source contributions (preferred).
  • Experience with LLM pre-training, post-training methods such as SFT, RLHF, or DPO, GPU performance work, efficient inference, cluster infrastructure, or large-scale data engineering (preferred).

Interested in this position?

Apply directly on the company website

Apply Now

Similar Roles

Research Scientist - Generative Audio

Spotify Media

Join Spotify’s Artist-First AI Music Lab as a Research Scientist to advance generative audio technologies that create new artist-centered music and listening experiences.

Computer Vision Machine Learning NumPy Python PyTorch Reinforcement Learning
3 hours, 42 minutes ago

Research Scientist - Generative Audio

Spotify Media

Spotify is hiring Research Scientists at all seniority levels for its Artist-First AI Music Lab to advance generative audio technologies and create new artist-centered listening experiences.

Computer Vision Generative AI Machine Learning NumPy Python PyTorch Reinforcement Learning
3 hours, 42 minutes ago

Research Scientist, Text-to-Speech

Oddin.gg 51-200 Hotels, Restaurants & Leisure

Join Valka.ai as a TTS Researcher to develop and deploy high-quality, low-latency voice generation models for interactive digital content in entertainment and education.

AWS GPT Machine Learning Python PyTorch Transformers
2 days, 3 hours ago

Applied Researcher (Senior+)

Miris Internet Software & Services

Miris is seeking a senior Applied Researcher to advance neural-rendering technology and bring efficient, high-fidelity 3D streaming experiences to low-powered mobile AR/VR devices.

Augmented Reality Python Virtual Reality
2 days, 4 hours ago

You're on a roll! Sign up now to keep applying.

Sign Up

Already have an account? Log in

Used by 14,729+ remote workers