Join Mirendil as a Member of Technical Staff to shape pretraining data for AI models in a fast-paced startup environment.
Posted by employer 2 days ago
First seen on Joblaze 1 day ago
Last verified on the company career page 1 day ago
What you'll build
- Build data pipelines for pretraining
- Extract, filter, and deduplicate public data
- Decide data mixture for training runs
- Direct agents to run experiments and pipelines
- Maintain data corpus cleanliness
Must have
- Shipped a data change into a large pretraining run
- Built data pipelines at the scale of trillions of tokens
- Strong Python skills
- Experience with PyTorch or JAX
- Hands-on experience with deduplication or quality classifiers
Not disclosed in this posting: years of experience, work arrangement, visa sponsorship.
Benefits
Joblaze summary
In this role, the individual will focus on building and managing data pipelines for pretraining AI models, ensuring the quality and relevance of the training data. Proficiency in Python and experience with large-scale data processing frameworks like Spark or Ray are essential, along with a solid understanding of PyTorch or JAX for running training experiments. This position is suited for someone with a strong background in data engineering and a track record of implementing data changes in pretraining runs. The team is dedicated to advancing AI research by refining data strategies and methodologies.
Joblaze insights
- Listed yesterday — first seen on Joblaze November 10, 2026. Last confirmed on Mirendil's careers page November 10, 2026.
- Salary band is above the typical range for AI/ML roles (median ~$189,154).
- Starts above 80% of 88 comparable staff ai/ml roles in United States that list Python we track (median $230,000 across 36 companies). See Python salary trends
- Python appears in 46.7% of 270 comparable staff ai/ml roles in United States; Beam appears in 0.4% of 270 comparable staff ai/ml roles in United States.
Quick facts
- What's the salary range?
- Mirendil lists $300,000–$400,000 for this role.
- What's the tech stack?
- Joblaze extracted these technologies from the posting: Beam, JAX, PyTorch, Python, Ray, Spark.
- What seniority level is this role?
- Mirendil targets staff-level candidates for this position.
- Is this full-time or contract?
- Full-time for this Member of Technical Staff, Pretraining Data role at Mirendil.
From the original posting
The Role
We pretrain our own models and we are scaling to larger models now. Agents handle much of the engineering, but deciding what goes into training still depends on strong human judgment.
You will set the direction for what our models train on and use agents to run the ablations, pipelines, and analyses that test it. You will work closely with the researchers who own architecture and training runs.
What You’ll Do
Work on building data pipelines for pretraining, including repository-scale code and technical documents.
Get more out of public data through better extraction, filtering, and deduplication of web pages, open-source code, and public datasets.
Decide the data mixture for each run, and design the ablations and data scaling experiments that justify it.
Direct agents to run those experiments and pipelines in parallel, and decide which results to trust.
Extend a code-first mixture into scientific domains, starting from where our models fall short.
Keep the corpus clean as it grows: provenance, decontamination against our evals, and reproducible versions.
About You
You have shipped a data change into a large pretraining run and can show the ablation that justified it.
You have built data pipelines at the scale of trillions of tokens with Spark, Ray, Beam, or similar, and you write strong Python.
You know PyTorch or JAX well enough to run your own training experiments.
You have hands-on experience with at least two of: deduplication, quality classifiers, mixture optimization, data scaling laws, or long-context and code data.
You can tell when an experimental result is real, and you explain your reasoning clearly in writing.
If you’re excited about shaping what our models learn from, and proving it with experiments, we’d love to hear from you.
Standard company text repeated across Mirendil's postings is omitted here.