Principal Engineer, Generative AI
Unit21
full-time
Posted on:
Location: California • 🇺🇸 United States
Visit company websiteSalary
💰 $250,000 - $350,000 per year
Job Level
Lead
Tech Stack
AWSCloudGoogle Cloud PlatformPythonPyTorchRuby on Rails
About the role
- Own the technical vision for the GenAI platform including retrieval-augmented pipelines, vector-search, fine-tuning, and evaluation harnesses
- Architect and ship investigative agents that ingest terabytes of transactional and case data, reason over heterogeneous knowledge graphs, and generate fully auditable explanations
- Lead model strategy: select OSS vs commercial LLMs, run parameter-efficient tuning, implement safety and bias mitigations
- Design MLOps/GPU infrastructure for low-latency inference and predictable cost per investigation
- Introduce robust evaluation and guard-rails including automated red-teaming, toxicity filters, PII leak detection, and latency/cost dashboards
- Mentor a senior engineering team through design reviews, pair programming, and knowledge-sharing
- Partner with Product and Compliance to translate AML workflows and regulatory requirements into agent toolsets
- Prototype state-of-the-art techniques and assess fit for production
Requirements
- 10+ years software/ML experience
- 4+ years building deep-learning or NLP systems
- At least 2 years directly on LLM or GenAI products
- Track record of architecting production GenAI systems at scale in the cloud (GCP/AWS)
- Deep understanding of RAG, vector DBs, prompt-engineering patterns, and evaluation metrics
- Expert in Python with hands-on experience with PyTorch
- Demonstrated leadership as Staff+ IC or Tech-Lead Manager, guiding 65 engineers on ambiguous 0120 initiatives
- (Bonus) Prior exposure to fraud-tech, payments, or AML domains
- (Bonus) Published research / open-source contributions in LLM safety, retrieval, or agentic frameworks