Open to opportunities

Aashish Rana

@aashishrana

Message

Data Scientist delivering production LLM systems, RAG, and enterprise NLP.

India

Message

What I'm looking for

I’m looking for a role where I can build production LLM/RAG systems—hybrid search, agents, and scalable data/ETL pipelines—owning end-to-end delivery and continuously improving accuracy with rigorous evaluation.

I’m a Data Scientist with 3+ years of hands-on experience building production AI systems for enterprise clients. I architect LLM solutions like Retrieval-Augmented Generation (RAG), AI agents, and LLM-powered chatbots that move from prototype to deployment.

At Tatras Data, I designed production-grade RAG pipelines using Pinecone and AWS Bedrock, combining semantic + keyword hybrid search with reranking to improve response accuracy. I also built scalable LLM report generation pipelines with AWS Bedrock and OpenAI APIs.

I focus on practical, end-to-end delivery across the AI and data stack—building SQL agents and GraphRAG-based assistants, and developing ETL pipelines with PySpark, AWS Glue, and Apache Airflow into Snowflake for downstream analytics. I’ve also implemented summarization pipelines using LLMs and FAISS.

I’ve strengthened core ML/NLP performance through fine-tuning transformer models like BERT and RoBERTa for multi-label and multi-class classification, and I’ve led teams by mentoring interns to design and deploy an AI-powered research paper chatbot. Earlier, as a Junior Data Scientist and Intern, I built machine learning pipelines and worked with Sentinel-2 imagery for object detection and land-use classification.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

Tatras Data

May 2023 - Present (3 years)

Built RAG-based LLM systems (Pinecone, AWS Bedrock) with hybrid search and reranking. Developed LLM report pipelines (Bedrock, OpenAI). Created SQL agents and chatbots on OpenSearch (millions of rows). Built ETL pipelines (PySpark, Glue, Airflow → Snowflake). Implemented summarization (FAISS) and fine-tuned BERT/RoBERTa. Led AI research chatbot team.

Python Generative AI NLP Machine Learning Machine Learning Pipelines Deep Learning AWS OpenSearch Pinecone Vector Databases

Jr. Data Scientist

Tatras Data

Sep 2022 - May 2023 (8 months)

Built end-to-end ML pipelines (preprocessing, feature engineering, training, evaluation) using Python, scikit-learn, and PyTorch. Fine-tuned BERT/RoBERTa for multi-class text classification on domain datasets. Mentored interns on satellite imagery classification projects.

Python Generative AI NLP Machine Learning Machine Learning Pipelines Deep Learning AWS OpenSearch Pinecone Vector Databases

Data Science Intern

Sabudh Foundation

Jan 2022 - Jul 2022 (6 months)

Built an object detection and land-use classification system using Sentinel-2 satellite imagery by extracting and processing multi-temporal data from Sentinel Hub with cloud filtering, noise reduction, and normalization. Trained and evaluated Random Forest and XGBoost classification models across datasets.

Satellite Imagery Analysis Object Detection Land Use Classification Sentinel 2 Data Normalization Random Forest XGBoost