Member of Technical Staff, Pretraining

HarkSan Jose AI Foundation Models

Apply on company site

Opens the official Hark careers page

Salary

$180K - $450K USD

Job type

Full-time

Posted

24 Mar 2026

Status

Open · checked 9d ago

About this role

Role Overview

The Member of Technical Staff, Pretraining position at Hark is an on-site role based in San Jose, focusing on the development of advanced foundation models and large-scale training systems. This position suits experienced AI engineers and researchers looking to shape the future of proactive, multimodal intelligence.

Key Responsibilities

Drive research and development within large-scale LLM and multimodal pretraining to enhance model capabilities.

Create and optimize data pipelines that handle curation, filtering, deduplication, and synthetic data generation.

Design and implement efficient training strategies, covering distributed training, scaling laws, and optimization techniques.

Build and improve core pretraining infrastructure, training systems, and compute efficiency.

Create evaluation frameworks and internal benchmarks to track pretraining progress.

Collaborate closely with cross-functional research and engineering teams to advance foundation model performance.

Requirements

Demonstrated track record of improving large-scale neural network performance through pretraining data, modeling, or training systems.

Extensive experience utilizing large-scale distributed training frameworks such as Megatron or DeepSpeed.

Comprehensive understanding of LLM or multimodal pretraining, data pipelines, scaling behavior, and optimization.

Practical experience in data-driven experimentation, systematic analysis, and debugging at scale.

Background in building or operating large-scale training infrastructure and high-performance computing systems.

Strong ownership mindset and ability to thrive in a fast-paced, research-driven environment.

Nice to Have

Prior experience working with multimodal pretraining across text, audio, and vision.

Skills and Tools

Large Language Models (LLMs)

Megatron

DeepSpeed

Distributed Training

High-Performance Computing Systems

Salary and Benefits

The annual base salary range for this position spans from $180,000 to $450,000 in the US. Actual pay varies based on individual factors like job-related knowledge, skills, and experience, with additional components or benefits potentially included.

About the Company

Hark builds advanced, personalized artificial intelligence designed to be proactive and multimodal. The company pairs this intelligence with next-generation hardware to establish a seamless human-machine interface that goes beyond traditional chat boxes.

How to Apply

Candidates should visit the official company careers page at hark.com to submit an application for this position.

Skills and tech stack

LLM

Questions candidates ask

Where is this job located?

This position is located in San Jose, California.

What is the salary range for this role?

The US base salary range spans from $180,000 to $450,000 annually, depending on experience and qualifications.

What team will I be working with?

You will be working with the Omni team at Hark, which focuses on building next-generation AI experiences beyond text.

What skills are required for distributed training?

Applicants need strong experience with large-scale distributed training frameworks such as Megatron or DeepSpeed.

Is multimodal experience required?

Multimodal pretraining experience involving text, audio, and vision is listed as a strong plus rather than a strict requirement.

Ready to apply?

The application is on the official Hark careers page.

Apply on company site