SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Sciforium is an AI infrastructure company developing next-generation multimodal AI models and a proprietary high-efficiency serving platform. Backed by multi-million-dollar funding and direct sponsorship from AMD, the company is scaling rapidly to build the full stack powering frontier AI models and real-time applications.
You will lead the strategy, creation, and curation of massive datasets that power foundation models. This role owns the end-to-end data lifecycle—from raw web-scale crawling to fine-grained human-alignment datasets that define model behavior. You will design taxonomies, filtering heuristics, and post-training pipelines to ensure models excel in reasoning, safety, and multimodal understanding.
Key responsibilities include:
- Owning end-to-end pre-training dataset creation for LLMs, defining the optimal mix of web data, code, books, and technical papers
- Designing and implementing sophisticated pipelines for data cleaning, deduplication, and high-quality signal extraction from petabytes of raw, unstructured data
- Leading development of high-quality post-training datasets including Supervised Fine-Tuning (SFT) instructions, multi-turn dialogues, and preference modeling data (RLHF/DPO)
- Driving acquisition and processing of vision and video data, navigating multimodal alignment, video compression, and temporal data consistency
- Developing high-throughput data processing scripts using Python with multiprocessing and multithreading for massive-scale ingestion and transformation
- Conducting deep-dive statistical analysis on training corpora to identify biases, gaps, and quality regressions
- Designing pipelines to generate high-reasoning synthetic data to augment natural datasets
Requirements:
- 5+ years of industry experience in Data Science or Machine Learning with proven track record building and managing datasets for foundation models
- Expert-level Python proficiency with focus on high-performance code, multiprocessing, multithreading, and efficient memory management for large-scale data tasks
- Demonstrated experience working with petabyte-scale datasets directly used to train production-grade LLMs or Large Vision Models
- Experience building massive LLM training sets from scratch, including raw web crawls (e.g., Common Crawl) and specialized domain data
- Hands-on experience building datasets for RLHF, DPO, and multi-turn instruction following, including management of human-labeling workflows and quality gold-sets
- Mastery of data-at-scale frameworks such as Spark, Ray, or high-performance data-loading formats (e.g., WebDataset, Parquet)
Nice-to-haves:
- Computer Vision (CV) curation experience building large-scale image or video datasets from scratch (e.g., LAION-style pipelines)
- Familiarity with large-scale multimodal crawling and associated challenges of video processing, codecs, and compression
- Experience designing complex labeling schemas for reasoning, coding, and mathematical benchmarks
- Master's or PhD in a quantitative field with focus on data-centric AI or information retrieval
