Open to opportunities

Sai Chandra Kiran Teja Podili

@saichandrakirantejap

Message

Data Scientist building evaluated RAG systems and fraud models with rigorous pipelines.

United States

Message

What I'm looking for

I’m looking for a Data Scientist role where careful modeling, honest evaluation, and clean data pipelines matter—especially building and tuning systems end-to-end with measurable quality, traceability, and reliability.

I’m a Data Scientist with a clear growth path—from an associate on a fraud analytics team at DBS Bank to building evaluated ML solutions as a Data Scientist at Docusign. I build ML models and feature pipelines from scratch, then back them with careful tracking and explainability so teams can trust what the system is doing.

Most recently, I contributed to the data and evaluation side of a production RAG pipeline at Docusign, focusing on ingestion, embedding, and retrieval quality—not frontend or infrastructure. I owned semantic chunking and embedding with OpenAI embeddings, wrote data to Pinecone and PostgreSQL with pgvector, streamed events through Kafka MSK, and used Ragas faithfulness and context recall scores with LangSmith traces to tune chunk size/overlap until faithfulness stabilized above 0.85.

Experience

Work history, roles, and key accomplishments

Data Scientist

DocuSign

Feb 2025 - Apr 2026 (1 year 2 months)

Owned the data ingestion, embedding, and retrieval evaluation for a production RAG pipeline for legal document processing, including semantic chunking, OpenAI embeddings, and storage in Pinecone and PostgreSQL/pgvector. Improved retrieval quality by tuning chunking until Ragas faithfulness stabilized above 0.85 and reduced repeat API calls via Redis caching, while tracing LLM calls end-to-end in L

Embeddings Pinecone Pgvector Ragas MLflow Experiment Tracking

Associate Data Scientist

DBS Bank

Aug 2022 - Nov 2023 (1 year 3 months)

Built an XGBoost fraud classifier for a 1:200 fraud dataset using SMOTE and tuned the decision threshold with precision-recall calibration to match the risk team’s tolerance. Engineered transaction features in PySpark and used SHAP for compliance-ready explanations, with experiments tracked in MLflow and models promoted to SageMaker endpoints.

XGBoost SMOTE SHAP Explainability Pyspark Fraud Models MLflow Experiment Tracking Risk Scoring

Education

Degrees, certifications, and relevant coursework

University of Alabama at Birmingham

Master of Science, Data Science

2024 - 2025

Activities and societies: MS capstone: Anime Face Generation using DCGAN (PyTorch/NumPy; GPU training via Google Colab), tracking FID and stabilizing adversarial training with label smoothing and batch normalization.

Master of Science in Data Science covering machine learning, deep learning, NLP, big data analytics, and statistical modeling. Completed an MS capstone on DCGAN-based anime face generation and tracked FID to evaluate training quality.