Projects

Detailed case studies of ML systems I’ve built. Each project includes the problem statement, approach, results, and lessons learned.

Featured Projects

Built a scalable NLP system for classifying customer feedback across batch and real-time pipelines.

Problem: Scalable text classification system to help organizations prioritize and respond to customer feedback at scale

Tech Stack: PySpark, HuggingFace Transformers, Databricks, MLflow

Impact: Deployed batch and real-time pipelines processing large-scale datasets with MLflow tracking and automated retraining

Built an ensemble forecasting system combining LSTM and tree-based models for accurate hourly climate prediction.

Problem: Accurate hourly climate prediction combining multiple modeling approaches.

Tech Stack: LSTM, Random Forest, Bayesian Optimization

Impact: 92% error reduction over baseline through ensemble methods.

Additional work exploring ML and optimization

SelfCurate (ongoing): A Python pipeline for iterative dataset curation and self-improving model training
Renewable Energy Optimization (ongoing): Reinforcement learning to handle uncertainties in renewable energy flux
SprachenKarte: An Interactive Language Diversity Visualization Tool

Model Compression & Quantization: Techniques for making models 10x faster with minimal accuracy loss
Rust for Data Processing: Why high-performance languages matter for ML infrastructure
Causal Inference: Moving beyond correlation to understand actual cause-and-effect in experiments
LLM Fine-Tuning: Practical approaches to adapting large language models for specific domains

Last updated: April 2026

Want to see more? Browse the blog for deep dives on specific techniques and lessons learned.