Narayan Nandeda
Verified Expert in Engineering
Machine Learning Developer
Narayan is a data scientist with 10+ years of experience in statistical modeling, NLP, ML, deep learning, AI, and GenAI. He's experienced in deploying ML solutions across the healthcare, retail, supply chain, telecom, and eCommerce domains. Narayan has been designing, developing, and deploying data science solutions using Python and R since 2011. He's an expert in supervised and unsupervised ML, regression, classification, forecasting, reinforcement learning, GANs, and Keras.
Portfolio
Experience
Availability
Preferred Environment
Jupyter, R, Python, Windows, Artificial Intelligence (AI), Classification Algorithms
The most amazing...
...project I've implemented is "Probabilistic Graphical Models(PGM)." I've also built "Insights Engine" for a supply chain company.
Work Experience
Lead Data Scientist
Blue Yonder
- Developed and deployed an ML model to predict the length of stay (LOS) of patients in the emergency department of the hospital. Utilized BERT, ALBERT, and Roberta models along with supervised ML models for this task.
- Developed generative adversarial network models to generate synthetic tabular data of the healthcare records like lab test values, vital signals, age, gender, and the like. Utilized GAN architectures, CTGAN, and more for this task.
- Built a COVID-19 mortality prediction model to predict the mortality of the disease in positive patients from day 10.
- Developed ML/DL models to predict the "Discharge/Admit" status of the patients who come to the emergency department of the hospital.
- Built and deployed GCP (CPU/GPU) based ML models and built a GCP infrastructure to use AI Notebooks.
Senior Data Scientist
Blue Yonder
- Converted LP formulation into an image and used computer vision techniques to decompose large LP problems into small subproblems.
- Created a deep learning autoencoder pipeline to convert the supply chain to a fixed-dimension vector.
- Used page ranking to relatively rank supply chain exceptions. Used historical click-stream data along with exception properties to learn the patterns and importance.
- Built probabilistic graphical models (PGM) to create an insight engine for a supply chain client. PGMs were designed to help planners to automate planning and decisions.
- Developed time series forecasting models to forecast daily inbound and outbound volume of 100+ distribution centers of a retail client. Utilized ARIMA, SARIMA, ETS, and PROPHET and supervised ML models for them.
- Developed a forecasting model for a retail client to forecast the journey time of trailers to arrive at stores.
- Created a sales forecasting model for a retail client to forecast daily sales at different granularities like daily, weekly, store-level, store-department-level, and more.
Data Scientist
Verizon Data Services
- Implemented real-time cancel propensity model to predict the cancel order propensity score for the customer in real time, before he submits the order.
- Implemented real-time churn model to predict the churn propensity of the customer in real time, which was deployed using the IBM info-streams platform.
- Worked on happy path scoring that takes the path the customer traverse through in the website and scores it as a happy path or unhappy path with a lift score to completion.
- Implemented session categorization and customer segmentation. The goal was to categorize the customers' sessions based on the activity they do in self-serve channels and use sessions to segment customers into clusters to generate complex insights.
- Implemented a next-day call prediction model to predict the propensity of call.
- Trained and deployed Call and Chat Classification model, to classify the transcript in various categories. Worked end to end starting from data collection to deployment (Call Classification and Chat Classification Model).
- Used NLTK to clean and preprocess text data, Stanford library for dependency parsing, BERT Model for embeddings, LDA for Topics identification, and more (Call Summary model).
- Implemented a sales forecasting model to forecast weekly sales for each store. Utilized Time series algorithms like ARIMA, SARIMA, Exponential Smoothing, Supervised ML Algo, and Deep Learning like LSTM, RNN (Sales Forecasting at Store Level).
Business Analyst
Verizon Data Services
- Developed a model to calculate customer lifetime values based on the profile and transactional data.
- Did insights generation using state-of-the-art ML algorithms for the representative to better serve customers.
- Implemented customer segmentation using online transaction details.
- Implemented Next Best Offer model (NBO). The NBO model identifies the next offer that can be given to a customer from all existing offers. It makes sure that the conversion rate increases.
- Implemented an agent-customer mapping model to improve call center agents' performance.
Experience
Generative Adversarial Network (GANs)
Computer Vision to Identify Scratches on The Surface of Items in Manufacturing
The objective was to have an automated quality check using computer vision.
Albert, Bert, and GPT2 Language Models
Computer Vision to Identify Physical Defects on Phone Bodies
Optical Character Recognition
I am also aware of AWS Textract.
Education
Master of Business Administration (MBA) Degree in Business Analytics
Indian Institute of Management Indore (IIM - Indore) - Indore, India
Bachelor of Engineering Degree in Computer Science and Engineering
Rajiv Gandhi Technical University, (RGPV- Bhopal) - Bhopal, India
Certifications
Machine Learning (Statistics and Machine Learning Micro Master)
MITx
Statistical Learning
Stanford University
Reinforcement Learning
National Research University — Higher School of Economics
Bayesian Methods for Machine Learning
National Research University — Higher School of Economics
Deep Learning Specialization
deeplearning.ai
Data Science Associate
Dell EMC
Machine Learning
Stanford University School of Engineering
Oracle Certified Associate (OCA)
Oracle
Skills
Libraries/APIs
Natural Language Toolkit (NLTK), XGBoost, Keras, OpenCV, NumPy, Pandas, Scikit-learn, SciPy, TensorFlow, PyTorch, TensorFlow Deep Learning Library (TFLearn), Matplotlib, LSTM
Tools
Jupyter, Google AI Platform, ChatGPT, Tableau, Plotly
Languages
Python, R, SQL
Paradigms
Linear Programming, Data Science
Platforms
Google Cloud Platform (GCP), Jupyter Notebook, Docker, Linux, Windows
Industry Expertise
Healthcare, Telecommunications
Storage
Google Cloud, Data Validation
Frameworks
Flask, RStudio Shiny
Other
Data Mining, Neural Networks, Statistics, R Programming, Statistical Modeling, Deep Learning, Linear Regression, Machine Learning, Decision Trees, Random Forests, Natural Language Processing (NLP), Mathematics, Clustering Algorithms, Classification Algorithms, Principal Component Analysis (PCA), BERT, Google BigQuery, Convolutional Neural Networks (CNN), LSTM Networks, Recurrent Neural Networks (RNNs), R-CNN, Hugging Face Transformers, GPT-2, Computer Vision, Image Processing, Predictive Analytics, Statistical Analysis, Data Analysis, Data Analytics, Data Visualization, Data Architecture, Artificial Intelligence (AI), Supply Chain, Supply Chain Optimization, Data Cleaning, Probabilistic Graphical Models, Generalized Linear Model (GLM), Computer Vision Algorithms, Google Cloud Machine Learning, Exploratory Data Analysis, Naive Bayes, Bayesian Inference & Modeling, Hugging Face, GPT, Generative Pre-trained Transformers (GPT), Language Models, OpenAI GPT-3 API, OpenAI GPT-4 API, Large Language Models (LLMs), XLNet, RoBERTa, Generative Adversarial Networks (GANs), OCR, Big Data, Deep Reinforcement Learning, Probability Theory, Linear Algebra, Linear Optimization, Containers, EMC Certified Data Scientist, Dash, Clustering, Demand Sizing & Segmentation, Modeling, Logistic Regression, Forecasting, Regression, Random Forest Regression, Long Short-term Memory (LSTM), Autoregressive Integrated Moving Average (ARIMA), Health IT
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring