About us Black Ore is building the leading AI platform designed to accelerate financial services. The company's flagship offering, Tax Autopilot, leverages proprietary AI agents to automate tax preparation and compliance, helping CPA & tax firms overcome the talent shortage and accelerate growth for their practice. Founded in 2022, we launched with $60 million in early stage funding from some of the world’s leading investors including a16z, Founders Fund, General Catalyst, Khosla Ventures, Oak HC/FT, Trust Ventures and leading tech founders/angel investors including Jason Gardner (Founder and CEO of Marqeta), Max Levchin (Founder of Paypal and Affirm), Tom Glocer (Former CEO of Thomson Reuters), Gokul Rajaram, and Mark Britto (EVP, CPO, PayPal). Our team has an incredibly ambitious vision to completely transform the way businesses and consumers interact in financial services. We’re looking to hire strong team members to grow the team. Some of the traits we look for are: Owner Mentality - Desire to take initiative, identify problems, and implement solutions Mission Driven - Passion for building AI/ML solutions that reimagine how businesses and consumers operate Intellectually Curious - Excitement going deep for building detailed understanding of the function, role, customer, and problem space Team Oriented - Ability to collaborate respectfully and put the team above the self What We Offer Competitive salary and equity based compensation Best-in-Class health benefits Platinum, 100% Employer-paid medical, dental and vision insurance Health Savings Account (HSA) and Flexible Spending Account (FSA) Additional Health programs (e.g., One Medical, Talkspace, Kindbody) 401K, Roth 401k and other employer sponsored investment benefits Parental Leave Unlimited PTO Relocation support to Austin, NYC or SF (as needed) WFH stipend to support your home office needs Qualifications Required 8–12+ years building and shipping applied ML / NLP / LLM systems in production. Strong experience in summarization, classification, extraction, and NER. Direct experience training, fine-tuning, evaluating, and deploying LLMs. Expert Python + PyTorch; able to own full ML pipelines independently. End-to-end production ML experience (data → modeling → infra → deployment → monitoring). Ability to move fast, iterate tightly, and ship working systems in resource-constrained environments. Strong debugging ability across the stack (data quality, tokenization, model behavior, infrastructure). Preferred Prior experience in document intelligence, financial systems, or enterprise automation. Experience building multi-model or multi-agent architectures. Experience designing evaluation frameworks for high-stakes workflows. Experience operating in a startup or 0→1 environment. The Role We are looking for a hands-on Applied AI Leader who will own the design, training, and deployment of our core AI systems. This role is execution-heavy: building models, writing code, designing pipelines, debugging failures, and shipping production AI. You will directly build and scale LLM-based extraction, classification, summarization, reasoning, and agentic systems that power the end-to-end Tax Autopilot workflow. If you want to build, train, fine-tune, evaluate, and ship—this role is for you. What You Will Do Model Development (Hands-On) Design and implement NLP/LLM systems for extraction, summarization, classification, and NER Fine-tune, distill, and optimize LLMs for tax-domain tasks. Build evaluation frameworks, datasets, and automated testing pipelines. Implement prompt engineering, retrieval strategies, and agent workflows. Production Engineering Own end-to-end model deployment into production environments. Build and maintain training pipelines, inference services, and monitoring systems. Debug real-world performance issues across data, models, retrieval, and orchestration. Drive continual improvement through tight iteration loops on accuracy, speed, and cost. Execution & Ownership Translate ambiguous operational and product problems into concrete ML approaches. Ship production-ready Python and service code—no hand-off to others. Work directly with engineering to integrate models deeply into application workflows. Prioritize high-impact improvements and cut what doesn’t matter.