Peter is a Senior Machine Learning Engineer with 6 years of experience in software development and 4 years of ML expertise, contributing to various open-source projects like PyTorch, XGBoost, and TensorFlow. He delivers Machine Learning infrastructure using Terraform and optimizes models for faster inference speed and lower memory usage, ensuring efficient utilization of computational resources. With hands-on back-end development experience in Python, Peter deploys packages and actively participates in AI, NLP, and CV research. He also has a strong background in cloud computing and MLOps and is a three-time certified AWS professional.
Working as the main contributor/designer of the Prediction Market Agent Tooling library and the Prediction Market Agents themselves, who achieve 75% and above accuracy in the prediction of future events.
Writing contracts for prediction markets and publishing subgraphs.
Migrated an in-house data pipeline into Databricks (running on Spark and Delta Tables), reducing the runtime of processing 60M emails from days to hours and halving the costs.
Refactored 2 models to MLflow and deployed on Databricks’ serverless endpoints.
Fine-tuned and served GPT-3 models.
Wrote and debugged prompts for the best behavior from pre-trained one/few shot models.
Developed and deployed a Siamese neural network and XGBoost models for a product pairing system.
Managed the system's architecture, utilized Kafka for communication, and implemented MLflow in Kubernetes.
Ensembled models using FastText and boosting trees, including a training pipeline and automated data analysis for re-training, and deployed in production using Docker and Helm.
