Member of Technical Staff [AI/ML Engineer]
$150k - $275k • San Francisco, United States • full_time
Posted 1mo ago
Job Location
San Francisco, United States
Tech Stack
Remote Work Policy
On-site
Employment Type
full_time
Categories
Machine Learning Engineer
About the job
Burnt is building the Operating Brain for the global supply chain, a system designed to run businesses on autopilot. We transform historical data from ERP systems into actionable forecasts, ontologies for AI agents, and models tuned on proprietary data. Our initial focus is the food industry, a complex and operationally challenging sector. We are a highly competitive team dedicated to building the future of supply chain operations. This role is for our first dedicated ML hire, responsible for building and evolving the model layer of our system. Models must be production-ready, continuously improving, and cost-effective.
Responsibilities
- Own MLOps end-to-end, including model creation, deployment, iteration, monitoring, and support.
- Build and maintain time series forecasting models for production traffic and backtesting.
- Fine-tune LLMs using LoRA and PEFT on proprietary data, and develop evaluation metrics for model deployment.
- Design and maintain the ontology and knowledge graph for AI agent reasoning.
- Engineer data pipelines at scale using Spark for messy, multi-tenant supply chain data.
- Develop versioning, drift detection, and retraining pipelines for deployed models.
- Manage the AWS ML stack, including SageMaker, S3, Glue, and Step Functions.
- Architect production systems for models and own associated design decisions.
- Contribute to full-stack development, including APIs and user interfaces, when needed by the team.
Requirements
- Expert-level proficiency in Python.
- Experience with Apache Spark and the surrounding data engineering ecosystem.
- Familiarity with AWS SageMaker, S3, Glue, and Step Functions.
- Experience with MLOps tools like MLflow or Kubeflow.
- Experience fine-tuning LLMs with LoRA and PEFT libraries.
- Experience with time series forecasting libraries such as Prophet, NeuralForecast, or statsmodels.
- Familiarity with knowledge graph technologies like Neo4j, RDF, OWL, or SPARQL.
- Sufficient TypeScript and React knowledge to contribute to the application layer.
- Proven experience owning MLOps end-to-end.
- Experience fine-tuning LLMs on real-world datasets.
Benefits
- Equity