I work on large-scale machine learning systems, focusing on the design, training, and deployment of models that operate reliably under real-world conditions. My interests include large language models, multimodal architectures, retrieval-augmented generation, and the data/compute infrastructure required to support them.

My work spans the entire lifecycle of modern ML systems: dataset construction, training pipelines, evaluation methodology, and inference optimization. I care about clarity in system design, reproducibility, and empirical rigor. I also build tools that make model behavior more interpretable and controllable.

I hold a Bachelor's and Master’s in Computer Science (Machine Learning). I write about ML, systems, and experimentation at https://thenumbercrunch.com/.

Side projects include HiveHaven - a lightweight platform for international students seeking housing in the U.S., and PolNet - a data visualization tool for analyzing and visualizing U.S. congressional caucus memberships and political network data.

Current reading: “Build a Large Language Model (From Scratch)” by Sebastian Raschka.

Predictive Modeling, Large Language Models, Multimodal Models, Generative Modeling
Retrieval-Augmented Generation, Vector Search, Data-Centric Evaluation
Training Pipelines, MLOps, Distributed Systems, High-Throughput Inference

Javascript, Python, C/C++, C#, SQL
PyTorch, TensorFlow, Scikit-learn
LangChain, LangGraph, ElasticSearch, Neo4j
Apache Spark, Databricks, Hadoop (HDFS), Postgres, BigQuery
AWS SageMaker, Amazon Bedrock, Vertex AI, Azure ML
Docker, Kubernetes, Git, DVC

LinkedIn: https://www.linkedin.com/in/pathak-ash/
X: https://x.com/pathak_jeee
Email: ashutoshpathak@thenumbercrunch.com
Writing: https://thenumbercrunch.com/