H

Machine Learning Researcher - Mechanistic Interpretability & Context Compression

HHM Talent • United State
Visa Sponsorship
Apply
AI Summary

Machine Learning Researcher responsible for cutting-edge research in mechanistic interpretability and context compression, owning end-to-end projects from hypothesis development to production deployment. Must design experiments, train transformer models from scratch, build datasets and evaluation infrastructure, and deploy impactful models. Requires strong transformer expertise, large-scale training experience, and high agency in a fast-paced San Francisco startup environment.

Key Highlights
End-to-end ownership of LLM research projects from hypothesis to production
Cutting-edge work in mechanistic interpretability and context compression
Training transformer models from scratch on large GPU clusters
Key Responsibilities
Design and execute experiments in LLM context compression and mechanistic interpretability
Train transformer models from scratch, owning data, architecture, training loops, and evaluation
Build datasets, labeling pipelines, and evaluation infrastructure
Technical Skills Required
Transformer models Large-scale model training Mechanistic interpretability Context compression
Benefits & Perks
Equity
Visa Sponsorship
H-1B sponsorship
Nice to Have
Experience pretraining transformer models
Reinforcement learning or post-training experience for LLMs
Novel architecture or training method development with measurable results
Experience at frontier AI labs, leading university research groups, or early-stage AI startups
Exceptional technical achievements such as Kaggle, ICPC, IOI, ISEF, or similar competitions
Strong background in applied mathematics, computer science, or engineering

Job Description


HHM Talent is assisting a client in their search for a Machine Learning Researcher in San Francisco, CA.

Position Overview

The Machine Learning Researcher will own cutting-edge research in mechanistic interpretability, context compression, and transformer training. Researchers drive projects end-to-end-from developing hypotheses and curating datasets to training models on large GPU clusters, evaluating results, and deploying models into production. This is a highly autonomous research role where shipping impactful models matters more than publishing papers.

Responsibilities

  • Design and execute experiments in LLM context compression and mechanistic interpretability.
  • Train transformer models from scratch, owning data, architecture, training loops, and evaluation.
  • Build datasets, labeling pipelines, and evaluation infrastructure.
  • Research and prototype novel model architectures and training methods.
  • Deploy successful models into production to improve customer outcomes.
  • Read and reproduce current AI research while driving independent research initiatives.

Qualifications

  • Experience training machine learning models from scratch with full ownership of data, architecture, and training.
  • Strong understanding of transformers and modern deep learning techniques.
  • Hands-on experience with large-scale model training and experimentation.
  • Research mindset focused on rapid experimentation and production impact.
  • High agency and ability to self-direct research.
  • Willingness to work onsite in San Francisco in a fast-paced startup environment.

Preferred Qualifications

  • Experience pretraining transformer models.
  • Reinforcement learning or post-training experience for LLMs.
  • Novel architecture or training method development with measurable results.
  • Experience at frontier AI labs, leading university research groups, or early-stage AI startups.
  • Exceptional technical achievements such as Kaggle, ICPC, IOI, ISEF, or similar competitions.
  • Strong background in applied mathematics, computer science, or engineering.

Compensation And Benefits

  • $150,000 - $300,000 base salary depending on experience
  • Equity

Visa Sponsorship

  • H-1B sponsorship available through employer

Similar Jobs

Explore other opportunities that match your interests

Visa Sponsorship Relocation Remote
Job Type Full-time
Experience Level Not Applicable

capable

United State

Senior Python Developer (AI/ML)

Machine Learning
•
1d ago
Visa Sponsorship Relocation Remote
Job Type Full-time
Experience Level Not Applicable

IT Engagements, Inc.

United State
Visa Sponsorship Relocation Remote
Job Type Full-time
Experience Level Not Applicable

clera

United State

Subscribe our newsletter

New Things Will Always Update Regularly