Naga Pavan Nukala

adwij5@r.postjobfree.com +1-313-***-**** www.linkedin.com/in/nagapavannukala/

SKILLS

Programming Languages: SQL, MySQL, PostgreSQL, Redshift, Python, Numpy, Pandas, Scikit-learn, Tensorflow, Keras, R, AMPL, MATLAB, PySpark / Spark, Hadoop, Putty, Jupyter Notebook, PyCharm, Agile, JIRA, Qlik Sense, Snowflake, Tableau, MS Excel, JMP Pro, AWS (s3, glue, lambda), Google Apigee EDUCATION

University of Michigan - Dearborn M.S – Data Science Sep 2021 – April 2023

Great Learning PGP - Data Science Engineering

May 2018 – Oct 2018

Vellore Institute of Technology Bachelor of Technology – Engineering Jun 2014 – May 2018

PROFESSIONAL EXPERIENCE

Research Assistant University of Michigan-Dearborn Sep 2022 – Dec 2022

• Optimized selection of serverless functions from 350 repos availing NLP techniques (esprima parser, regular expressions, pattern matching) to detect library calls, API calls, and function calls, resulting in improved computational complexity. Senior Analyst Tredence Analytics

Mar 2019 – Jun 2021

• Guided a team of 4 analysts in crafting insightful decisions using statistical and data science methodologies, and effectively partnered with stakeholders to resolve business challenges.

• Deployed NLP and Machine Learning algorithms (BERT, LSTM, Random Forest) to efficiently de-duplicate customer entries for a personal care industry partner, delivering a 28% decrease in the marketing team's targeting budget.

• Lowered shipping costs by 20% by verifying customer and address information for a partner in 40 countries with a cutting-edge Python solution, integrating Google APIs and packages such as sklearn, nltk, gensim, and libpostal.

• Crafted a commercial AI solution with Explainable AI and Model Drift detection capabilities. Improved quarterly sales by 3% and reduced ML pipeline downtime from 4-6 days to 4 hours.

• Engineered a Predictive Analytics solution for a global building materials company to forecast Order Cancellations, saving 30% in labor costs during labor planning.

• Extracted data from Snowflake, Qlik sense and built Forecasting models to predict sales for stakeholders in 20 countries and crafting customized python dashboards (using “dash” package) to visualize KPIs such as historical & forecasted sales. Consultant Analyst Kantar Analytics

Oct 2018 – Feb 2019

• Analyzed campaigns for a retail sector client apply Marketing-Mix-Modelling techniques (Athena, R and MS Excel), improving client campaign strategy, and increasing KPIs (Top of Mind by 15%) over a 2-month period. CAPSTONE PROJECT University of Michigan Ann Arbor – ITS Dept Jan 2023 – Apr 2023

• Collaborated with University of Michigan Ann Arbor ITS to migrate 60 APIs from IBM to Google Apigee.

• Designed and deployed automated alerts using Python, Splunk, and GCP to keep stakeholders informed of critical events.

• Created comprehensive documentation for handled tasks to ensure seamless maintenance and future reference. ACADEMIC PROJECTS

Prediction of lead levels existence in water samples - Flint water crisis

• Categorized water samples by investigating the presence of high lead levels (above 15ppm) based on multiple classifiers

(kNN, Naïve Bayes, Ensemble models) in JMP Pro and compared AUC metrics to arrive at the best model.

• Visualized the relationships of geographic and demographic features for the data till March 2019 in Tableau. Recommendation systems for e-commerce platform

• Delivered recommendations based on item-item collaborative filtering for the Amazon electronics-user-product ratings datasets. Utilized python packages surprise, and scikit-learn to iterate and develop a model of reduced RMSE 1.34 Evaluation of Map- Reduce performance for multiple bigdata problem statements

• Evaluated Hadoop and PySpark systems in an HDFS environment for multiple use-cases (5-gram token count, item recommendation and matrix multiplication, etc.,) to evaluate the accuracy, performance for multiple datasets. Evaluation of CLIP (Contrastive Language Image Pre-training)

• Optimized OpenAI's CLIP, a cutting-edge neural network model, to learn image representations from natural language and leveraged it to generate NSFW filters and theme-based images for 200 image-caption pairs using Tensorflow. Identification of profitable customer segments for a Bank’s Personal loan campaign

• Segmented 10,000 bank customers with help of ML algorithms (Decision Trees and CART) in python packages Numpy, Pandas, scikit-learn, Matplotlib, plotly and leverage the classifier's prediction probability to minimize targeting cost. Prediction of Cardio-vascular issues

• Utilized Logistic Regression to forecast cardio-vascular problems for 500 patients, leveraging data from UCI repo. Also, performed comprehensive EDA to uncover relationships among predictors in R and Python. Patronizing Condescending Language Detection

• Designed and delivered 85% accurate solution for SemEval 2021 patronizing language detection using Naive Bayes and Deep Learning models with custom-trained word2vec features.

• Pre-processed text, including correcting spell mistakes and handling punctuations, to improve solution accuracy. AWARDS

Innovative Data Science product

• Received Innovative Data Science solution as a team - AI Data cleanser (AIDC) by a famous magazine in India. https://analyticsindiamag.com/13-leading-data-science-products-from-india-that-made-it-big-in-2019/ Quick off the block award

• Recognition award for leading and executing the project - Q1 2020, Tredence Analytics

Contact this candidate