ModalNew York, NYBuild and optimize ML training systems for production-scale language models using PyTorch and frameworks like Hugging Face. Requires 5+ years experience in high-performance code and training optimizations; onsite in NYC or SF.
Machine Learning Engineer owning the full ML lifecycle for multimodal video datasets at Sieve. Fine-tune VLMs, build evaluation/QA pipelines with frontier models, design filtering systems over internet-scale data, and ship production improvements for top AI labs. Requires strong Python, PyTorch, and production ML experience.
Member of Technical Staff conducting hands-on LLM inference research at Modal. Own end-to-end bets on techniques like speculative decoding, quantization, KV-cache management, and disaggregation to improve cost per token and tail latency on production workloads. Requires strong LLM serving stack expertise and a track record shipping research or systems.
Own the training pipeline for search and agent models, building from product usage data through fine-tuning and evaluation to production deployment. Requires deep expertise in transformer fine-tuning, data curation, and training models for ranking, retrieval, and agent behavior.
Own the multi-stage ranking pipeline for web-scale search, balancing precision, recall, latency, and compute cost across retrieval, first-pass ranking, and neural reranking.
Staff engineer builds AI-driven platform infrastructure for SDLC transformation, owns end-to-end experiments using AI agents like Claude, and ensures high-velocity code delivery with strong abstractions and real-world grounding. Requires staff-level architecture experience and AI-native workflows.