Who We Are:

We build infrastructure that delivers massive amounts of web data to the companies training the world’s most powerful AI models.

We're the team that helps to power and support Grass, a bandwidth-sharing network that lets us operate a massive distributed crawler, giving us unique access to high-quality public web data at global scale. On top of that, we’ve built pipelines for ingesting, segmenting, and annotating billions of videos, transcripts, and audio files, powering dataset creation for frontier labs.

We’re lean, technical, and move fast. No red tape, no slow decision-making; just a team of builders pushing to expand what’s possible for open web data and AI.

Overview:
As a Research Crawling Engineer, you will design and operate large-scale web data acquisition systems for research and model development. Your work will span distributed systems, scraping infrastructure, and data pipelines.

Please note: This role requires a work schedule that sufficiently overlaps with EST business hours to collaborate effectively with the team.

Responsibilities:

• Build and maintain large-scale web crawlers across diverse domains
• Design high-throughput, fault-tolerant systems for data collection (millions to billions of URLs/day)
• Handle anti-bot systems, rate limits, and dynamic/JS-heavy sites
• Develop pipelines for cleaning, deduplication, filtering, and normalization
• Construct and maintain datasets for research and model training
• Monitor crawl performance, coverage, and data quality; iterate quickly
• Collaborate with research teams to align data collection with modeling needs
• Optimize infrastructure for cost, latency, and reliability

Requirements:

• Strong programming experience in one or more of: Go, Rust, Python, Java, or C++
• Experience building web crawlers or large-scale data pipelines
• Solid understanding of HTTP, networking, and browser behavior
• Familiarity with distributed systems and parallel processing
• Experience working with large datasets (TB–PB scale preferred)
• Ability to debug unstable or adversarial environments

Preferred / Bonus:

• Experience with NLP pipelines or dataset curation for ML
• Familiarity with LLM pretraining data or retrieval systems
• Experience with headless browsers (e.g., Chrome DevTools Protocol, Playwright, Puppeteer)
• Knowledge of proxy systems, IP rotation, and large-scale request orchestration
• Background in data quality evaluation or benchmarking
• Experience running workloads on cloud or bare-metal infrastructure

What This Role Involves:

• Operating at the boundary of scale and reliability
• Adapting to constantly changing web environments
• Balancing throughput, coverage, and data quality
• Owning end-to-end data acquisition pipelines

Evaluation Criteria:

• Ability to design systems that scale without degrading quality
• Practical problem-solving under real-world constraints
• Speed of iteration and ownership
• Measurable improvements in data coverage, quality, or efficiency

Compensation:

Based on experience and demonstrated ability to operate at scale

Example Projects:

• Build a distributed crawler for a continuously updated, high-quality web project
• Design a system to classify and filter billions of pages for pretraining
• Extract structured data from dynamic, JS-heavy sites at scale
• Improve deduplication and quality scoring across multimodal datasets

Why Work With Us:

• Opportunity. We are at the forefront of developing a web-scale crawler and knowledge graph that improves access to public web data and extends the value of AI to the people.
• Culture. We're a lean team with a high bar. We come to work not to be comfortable, but to find out what we're capable of and to do work that matters. We're not calling for people who keep things moving. We're calling for people who make everyone around them better.
  We prioritize low ego and high output. This is a fully remote team.
• Compensation. You’ll receive a competitive salary & benefits.