Senior ML Engineer

ABOUT THIS POSITION Job Description Summary We are seeking a highly skilled and innovative Machine Learning Engineer with a passion for building robust, efficient, and domain-specific AI systems using Language Models (LMs) and agentic architectures. As a core member of the team, you will be instrumental in developing the entire ML pipeline, from sophisticated data extraction techniques to fine-tuning specialized LMs and orchestrating their interactions within a multi-agent framework. This is a unique opportunity to apply state-of-the-art Generative AI and NLP techniques to a real-world, high-impact problem, leveraging the latest research in agentic AI and LMs to deliver economical and powerful solutions. WHAT YOU'LL DO + Data Pipeline & Knowledge Base Construction: + Design, implement, and optimize robust pipelines for ingesting, parsing, and extracting structured information from complex documents (leveraging OCR, document layout analysis, Named Entity Recognition (NER), and Relationship Extraction (RE)). + Develop rich, nested JSON schemas for representing structured data and ensure scalable storage + Generate and manage high-quality vector embeddings for efficient retrieval-augmented generation (RAG) within a Vector Database. + Language Model (LM) Development & Fine-tuning: + Research, select, and experiment with appropriate open-source Language Models (Large & Small) (e.g., Phi-3, Mistral, Llama, Nemotron-H families) for specialized tasks. + Design and execute efficient fine-tuning strategies (e.g., LoRA, QLoRA, full fine-tuning) on curated, domain-specific datasets to achieve precise performance for tasks like coverage determination, code lookups, and policy rule application. + Explore and implement knowledge distillation techniques to transfer capabilities from larger models to smaller, more efficient LMs. + Agentic System Design & Implementation: + Build and maintain the core agentic framework, including the orchestrator that intelligently routes queries and coordinates interactions between various specialized LM tools. + Develop and integrate

Created: 2025-12-04