Open to opportunities

Zankhana Mehta

@zankhanamehta

Message

I build production ML systems for drug discovery, fintech analytics, and large-scale network optimization.

United States

Message

What I'm looking for

I'm looking to build production machine learning and AI systems where I can improve real-world decisions through scalable data pipelines, model development, MLOps, and applied research.

I've built machine learning systems at Antares Therapeutics, improving ADME prediction R² by 12%, reducing validation RMSE by 15%, and supporting prioritization of 200+ compounds through automated monthly model retraining on AWS SageMaker.

Previously, I improved fintech forecasting and fraud detection at DeetsDigital and processed petabyte-scale telecom data at Reliance Jio. I also build production-grade recommendation, agentic analytics, and RAG applications using Python, PyTorch, AWS, LangGraph, Neo4j, and Docker.

Experience

Work history, roles, and key accomplishments

Machine Learning Co-op

Antares Therapeutics

Jan 2026 - Aug 2026 (7 months)

Benchmarked graph-based deep learning architectures for ADME property prediction, achieving a 12% R² improvement and enabling chemists to deprioritize 30% of low-confidence candidates. Built a self-improving hyperparameter optimization agent using Bayesian search and designed the orchestration layer for a production monthly retraining pipeline.

ChemProp graph neural networks Bayesian Optimization AWS Sagemaker IAM Infrastructure as Code Claude Code

Data Science Intern

DeetsDigital

Mar 2024 - Jun 2024 (3 months)

Increased Return on Marketing Investment by 20% by building interactive Power BI dashboards and running ANOVA and A/B testing. Improved revenue forecasting accuracy by 18% using time series modeling and strengthened high-risk transaction detection by 30% using unsupervised anomaly detection.

Power BI ANOVA A B Testing ARIMA Prophet Time Series Analysis Anomaly Detection

Machine Learning Intern

Reliance JIO

Jun 2023 - Dec 2023 (6 months)

Engineered scalable PySpark ETL pipelines processing 1-5 petabytes of spatial network data and applied PCA-based dimensionality reduction. Fine-tuned a pretrained Inception v3 model via transfer learning, reducing RSRP prediction error by 5 dBm and automating data ingestion workflows via Bash scripting.

PySpark ETL PCA Transfer Learning BASH Data Engineering

Education

Degrees, certifications, and relevant coursework

Northeastern University

Master of Science, Data Science

2024 -

Grade: 4.00/4.00

Pursuing a Master of Science in Data Science at Northeastern University's Khoury College of Computer Sciences, with coursework in machine learning, data mining, and MLOps.

Vellore Institute of Technology

Bachelor of Technology, Computer Science and Engineering

2020 - 2024

Grade: 8.93/10.00

Earned a Bachelor of Technology in Computer Science and Engineering from Vellore Institute of Technology, with coursework in data structures, algorithms, and database management.