Senior Data Scientist · Bay Area, CA

Building intelligent
data systems that
drive decisions.

AI-native analytics engineer with 4+ years shipping production ML pipelines, clinical NLP, and self-serve analytics platforms — currently at Johnson & Johnson.

View Projects GitHub ↗ Research ↗

Years Experience

Companies

18%

Stock-out Reduction at J&J

40%

Faster Time-to-Insight

About

I'm a Senior Data Scientist who lives at the intersection of machine learning engineering, clinical data, and scalable analytics infrastructure. My work at Johnson & Johnson involves building AI-driven forecasting systems for patient adherence, deploying BERT-based NLP over clinical notes, and delivering BI dashboards that translate model outputs into decisions executives can act on.

Before J&J, I led demand forecasting and inventory optimization at Orion Technolab — end-to-end, from ETL pipelines through LSTM models to production APIs. I hold an MS in Data Analytics from San Jose State University (Dec 2024) and am actively publishing research on privacy-preserving ML for healthcare.

I'm a daily user of AI coding tools (Claude, ChatGPT, Cursor) and am upskilling on RAG architectures, agentic frameworks, and LLM pipelines — looking for roles where clinical domain expertise meets the frontier of GenAI.

Location Bay Area, CA (Los Altos / Palo Alto)

Email bhavanb2000@gmail.com

Phone 408-581-4706

GitHub github.com/bhavanbk2

Research Google Scholar

Visa H1B sponsorship required

HIGH5 Strengths

1 Coach

2 Self-Believer

3 Storyteller

4 Commander

5 Deliverer

Independently certified · Oct 2025. Coach + Storyteller + Deliverer = someone who builds reliable systems and communicates their impact clearly to any audience.

Experience

Feb 2025 — Present

Johnson & Johnson

USA

Data Analyst / Senior Data Scientist

Built production ML models for clinical risk prediction and patient outcome analysis using Python, TensorFlow, and scikit-learn. Deployed BERT + Hugging Face NLP pipelines over unstructured clinical notes and regulatory documents, reducing manual review by 30%. Owned end-to-end data governance with GitHub CI/CD, Docker, and Kubernetes — GxP-aligned. Designed AWS QuickSight, Tableau, and Power BI dashboards tracking operational KPIs and model performance for cross-functional teams including Product, Operations, and Research.

↑15% prediction accuracy ↓30% manual review 30% faster decisions 3× dev speed via AI tools GxP / regulatory compliant

Feb 2024 — Jun 2024

Quantiphi

Massachusetts, USA

Data Analyst

Built scalable ETL pipelines and data transformation workflows using Python (Pandas, NumPy), SQL, and Apache Airflow. Developed self-serve Tableau dashboards for customer support and operational metrics, eliminating 50+ monthly ad-hoc requests. Conducted A/B testing and statistical analysis to evaluate RNN/LSTM forecasting models for time-series predictions.

35% pipeline efficiency 22% forecast accuracy Eliminated 50+ monthly reports

May 2020 — Dec 2022

Orion Technolab

India

Data Analyst / Data Scientist

Led end-to-end demand forecasting and inventory optimization — from ETL (Kafka, Airflow, SQL) through LSTM/ARIMA modeling to production REST APIs. Built NLP pipelines with SpaCy and NLTK for sentiment analysis over 1M+ customer reviews. Trained CNN and GAN models for product image defect detection. Deployed all systems with Flask/FastAPI, Docker, CI/CD, and CloudWatch monitoring.

20% forecast accuracy ↓15% stockouts 28% defect detection 35% faster deployments 40% less data prep time

Key Projects

🧬

Healthcare AI

Patient Adherence & Demand Forecasting

AI-driven system to predict patient adherence and optimize therapy demand across multiple regions for J&J. Integrated 250M+ structured records and 30M+ clinical text entries. Built BERT NLP for adverse event detection and deployed via AWS SageMaker with drift monitoring.

18%

Stock-out reduction

40%

Faster insights

Python TensorFlow BERT AWS Glue Snowflake SageMaker Tableau Power BI

📦

Retail AI

Intelligent Demand Forecasting & Inventory Optimization

End-to-end system replacing manual spreadsheet forecasting for retail clients. Built automated ETL, compared ARIMA/Prophet/XGBoost/LSTM models, and integrated NLP sentiment features from 1M+ customer reviews. Deployed as REST APIs with serverless infrastructure and live monitoring.

20%

Forecast accuracy

15%

Stockout reduction

LSTM XGBoost SpaCy AWS Glue FastAPI Docker CloudWatch

🔬

Research

Healthcare ML Research & Publication

Active peer-reviewed research on privacy-preserving ML for behavioral health signal detection (JMIR AI) and LSTM-GPT-4 integration for biomedical signal classification. Demonstrates frontier-level clinical AI work beyond production engineering.

Publications

JMIR

Target journal

GPT-4 LSTM Privacy-ML Biomedical NLP Signal Classification

🏭

Computer Vision

Product Quality Defect Detection

CNN and GAN-based visual inspection system to classify manufacturing defects. Used transfer learning (ResNet, EfficientNet) and GAN-generated synthetic samples to overcome limited defective image data. Deployed as a lightweight inference API for QC teams.

28%

Detection improvement

Real-time

API inference

CNN GAN ResNet TensorFlow Flask Docker

Skills & Stack

Languages

SQL (Expert) Python R PySpark Pandas / NumPy

ML & Deep Learning

TensorFlow PyTorch Scikit-learn Keras XGBoost LSTM MLflow

NLP & LLMs

BERT / HuggingFace Transformers SpaCy / NLTK GPT-4 RAG Agentic frameworks

Cloud & Data Platforms

AWS S3 / Glue Redshift SageMaker Lambda Snowflake Databricks BigQuery Azure

Data Engineering

dbt Apache Airflow Apache Kafka Spark ETL / ELT Data Modeling

BI & Visualization

Tableau Power BI AWS QuickSight Looker Sigma Amplitude

MLOps & DevOps

Docker Kubernetes CI/CD GitHub Actions FastAPI Flask CloudWatch

Databases

MySQL PostgreSQL MongoDB Redis Redshift

Education & Certifications

MS in Data Analytics

San Jose State University, CA

Dec 2024

BE in Computer Science

Anna University, Chennai

May 2021

Certifications

Azure Fabric Analytics Engineer Associate

AWS Data Engineer Associate

Azure Fundamentals

Research

JMIR AI Active

Privacy-preserving ML for behavioral health signal detection

Biomedical Signal Classification

LSTM + GPT-4 integration for biomedical signal classification

Building intelligentdata systems thatdrive decisions.