Janhavi Kulkarni

Graduate Student Research Assistant @ University of Cincinnati

About

I build ML systems that work on real data. Right now, I'm a Graduate Researcher at the University of Cincinnati, where I'm building an end-to-end structural health monitoring system for the Brent Spence Bridge, one of the busiest bridges in the US. That means real sensor data from 50+ channels, automated pipelines, LSTM-based anomaly detection, and a live dashboard used by Ohio DOT engineers every day. My electrical engineering background in signal processing, time-series modeling, and sensor data analysis is what makes this work. I understand the data at the hardware level, not just the model level. Before that I worked on healthcare AI, a CXR radiology report generation system, retina blood vessel segmentation with U-Net, and a medical chatbot using RAG and LangChain. I also have a published paper from ICDSA 2023 on disease classification using computer vision. What I love most is building things that actually ship, not just models that score well on benchmarks but systems that run in production and solve real problems. I'm currently looking for internships and full-time roles. Programming Languages: Python, MATLAB, JavaScript, HTML & CSS, SQL, CRBasic.AI & Machine Learning: Deep Learning, Neural Networks, Computer Vision, NLP, LLMs, RAG, Image Processing.Frameworks & Libraries: TensorFlow, PyTorch, Keras, OpenCV, LangChain, React.js, Flask, NumPy, Pandas, Scikit-learn, Matplotlib, Seaborn, SciPy.Databases: MySQL, MongoDB, FAISS, Pinecone.Cloud & DevOps: AWS, Docker, Git/GitHub, MLflow, CI/CD.Electrical & Signal Processing: DSP, Linear Systems, Sensor Data Analysis, Time-Series Modeling, Probability & Random Processes, Embedded Systems Fundamentals.

Country

United States

City

Cincinnati

Industry

Information Technology & Services

Skill

LangChain, LLaMA, Communication, Problem Solving, Large Language Models (LLM), Artificial Intelligence (AI), Digital Signal Processing, Machine Learning, Deep Learning, Digital Image Processing, Convolutional Neural Networks (CNN), Python (Programming Language), Probability, Scikit-Learn, Research Skills, MATLAB, Image Processing, Image Segmentation, Support Vector Machine (SVM)

Experience

Graduate Student Research Assistant

University of Cincinnati

2025-1 - Present · 1 yr 9 mos

Cincinnati, OH

Engineered infrastructure health monitoring pipeline processing 200K+ multivariate sensor records with automated ETL workflows including timestamp normalization, validation, and MySQL storage across 16 sensor channels. Developed LSTM autoencoder anomaly detection model to identify structural deterioration patterns, achieving 2x–8x lower reconstruction thresholds than a Feedforward baseline, better temporal pattern learning. Deployed containerized AI inference pipeline using Docker and implemented React + Flask dashboard for structural monitoring across 50+ bridge sensors.

Undergraduate Research Assistant

Shri Ramdeobaba College of Engineering and Management

2022-8 - 2023-2 · 7 mos

Nagpur, Maharashtra, India

The aim of this research was to reduce computational time and enhance accuracy by identifying images as either monkeypox skin or non-monkeypox skin. • Implemented Histogram of oriented gradients as feature extractor (HOG) and Support vector machines (SVM) as a classifier to help in detection of monkeypox disease with reduced complexity. • Proposed and a machine learning model for its feature of reduced computational time for monkeypox disease detection, utilizing a dataset of approximately 1,150 images (both affected and non-affected by MPOX) to ensure robustness. • Increased the model’s accuracy (94%), F1 score (95%), Recall (97%) and Precision (93%) using machine learning which shows exceptional model accuracy.