PythonPyTorchLLMsRAGMLflowVector DBsSpark

Role Briefing: Work on the Mosaic AI team at Databricks in Seattle to engineer customized open-weights LLM training, quantization, and retrieval-augmented generation (RAG) platforms for Fortune 500 enterprises.

Team Overview

Databricks combines data, analytics, and AI on a unified Lakehouse platform. The Mosaic AI team focuses on enabling organizations to build, fine-tune, and deploy bespoke generative models on proprietary enterprise data.

Key Responsibilities

Develop state-of-the-art fine-tuning recipes (LoRA, QLoRA, DPO, RLHF) integrated natively into Databricks Model Serving.

Implement highly accurate hybrid retrieval engines coupling sparse BM25 indexing with dense neural embeddings and rerankers.

Benchmark and optimize model latency using vLLM, TensorRT-LLM, and flash attention primitives.

Publish technical case studies and open-source contributions to MLflow and standard AI research repositories.

Requirements

MS or PhD in Computer Science, Data Science, or related field with focus on NLP/ML.

4+ years of industry experience deploying production machine learning models.

Deep expertise in PyTorch, Hugging Face Transformers, and distributed evaluation frameworks.

← Return to All Tech Positions