Pushan P

About Pushan P

Data Engineer with hands-on experience building cloud-native data platforms on Google…

Experience

1. Ai Engineer

   Berkshire Hathaway

   Feb 2024 — Present · US

   Built and deployed GPT-based and RAG-enabled AI systems using FAISS and Pinecone, improving contextual retrieval accuracy by 38% and reducing information latency across enterprise applications.• Designed end-to-end MLOps pipelines with MLflow, Airflow, and Kubernetes to automate model retraining, versioning, and deployment, cutting manual effort by 55% and ensuring reproducibility.• Fine-tuned transformer architectures (BERT, GPT-J) on PyTorch, reducing inference latency by 27% and lowering AWS SageMaker compute costs by 22% through model optimization.• Created data processing workflows in Spark and Databricks that shortened ETL cycles from two hours to fifteen minutes, enabling near-real-time model refresh and analytics.• Containerized FastAPI microservices with Docker, deployed on SageMaker endpoints, achieving 99.99% uptime while processing over one million daily inference requests across client platforms.• Integrated explainability tools such as SHAP and LIME into production dashboards, helping teams interpret predictions and maintain compliance with internal audit and governance standards.• Worked closely with cross-functional teams in Agile sprints to deliver AI features 30% faster, aligning technical deliverables with measurable business outcomes and user adoption metrics.

Education

• Kendriya Vidyalaya Arakkonam

  Class 12, PCMB

  2018 — 2019

• University of Missouri-Kansas City

  Bachelor's degree, Computer Science

• Kendriya Vidyalaya Arakkonam

  Class 10

  2016 — 2017