About
I am a seasoned data professional with over 5 years of hands-on experience in Data cleansing, Data Analytics, Machine Learning. I am passionate about leveraging data-driven insights to drive business decisions and solve complex challenges. Alongside job, I am also pursuing AMPBA (MBA in Business Analytics) from ISB Hyderabad in hybrid mode.
Associate Consultant, Analytics @ Eli Lilly
Developing data-driven advanced analytics solutions for one of the biggest Pharmaceutical companies of the world, Eli Lilly.
- Degree: AMPBA (MBA in Analytics), Computer Engineering
- Schools attended: ISB Hyderabad(AMPBA), KIIT(BTech)
- Interests: Data Science, ML, NLP
- Email: sudeshna_kundu_ampba2024s@isb.edu
- Current Location: Bangalore, Karnataka
Highlights
Here are some of the highlights of my academic and professional career and extra curricular activities.
Months of relevant experience
Fulltime corporate projects
Awards
Personal projects
Skills
Here is a snapshot of some of my skills at present.
Tools & Languages: Python, scikit-learn, PyTorch, PySpark, xgBoost, SQL, MS-Fabric, Tableau, AWS, Power-BI
Resume
Here is a brief overview of my academic and professional career.
Summary
Sudeshna Kundu
I have a strong technical background and a varied exprience across Advanced Analytics, Machine Learning, Statistical Modelling etc, which enables me to solve problems in a analytical and efficient manner. I excel in applying various forecasting methods, from statistical models to machine learning algorithms, addressing complex time series challenges such as demand forecasting and anomaly detection. Currently with my learings at ISB, I am transitioning towards Deep learning, Neural Networks and other advanced ML technique based roles.
Education
Advanced Management Programme in Business Analytics
Jun 2023 - Dec 2024
Indian School of Business, Hyderabad, Telengana
Pursuing AMPBA(MBA in Business Analytics) at ISB specifically focusing on Machine Learning(Supervised and Unsupervised), Deep Learning, Artificial Neural Networks.
Academic Projects:
- Topic Modelling by Unsupervised Machine Learning.
- Categorized topics from news articles based on unsupervised machine learning approaches (K-Means, Latent Dirichlet Allocation)
- Visualized those new articles post segregation Self Organizing Maps based their similarity to one another.
- Github Link: Link to Project
- Customer & Product segmentation.
- Conceived customers purchase behaviour, segmented the products based on how customers buy them and categorize them.
- Achieved visibility of how products are being bought and clustered together, so that clusters can become basis for recommendations.
- Github Link: Link to Project
- Stock Price Prediction using Sentiment Analysis of News Headlines.
- Scored sentiments of news headlines for a particular company and analysed stock price based on previous news headlines.
- Used regression to predict the upcoming stock prices.
- Published portal using Flask, deployed the model locally using Streamlit.
- Github Link: Link to Project
Bachelor of Engineering, Computer Science
2015 - 2019
KIIT, Bhubaneswar, Odisha
A proud alumni of KIIT CSE 2015-2019.
Academic Projects:
- TBA
Other Experiences
Software Engineering Intern
Jan 2019 - June 2019
GS Labs, Pune, Maharastra
- Developed automated alerting system by processing real-time telemetry feeds from IoT sensors of transport vehicle linked to AWS IOT core
- Built pipeline to process data streams using AWS lambda & Kinesis stream & store in AWS Dynamo DB; Used Flask based APIs for retrieval
Professional Experience
Associate Consultant, Analytics
Feb 2022 - Present
Eli Lilly & Company, Bangalore, Karnataka
- Developed a novel predictive solution for demand & capacity metrics, prevented unnecessary expense by $120K/year
- Trained regression model on the % drop-in rates of medical writing demands, predicted the same for next 2 quarters with 78% accuracy
- Comprehended capacity surplus , produced a dashboard to track metrics of hiring for scientists, editors, and medical analysts
- Added visibility of predictive analytics, reduced descriptive analysis time from 15 days to 2 hours
- Implemented unsupervised machine learning to segment patients & medicines, potential to cut manual surveying cost by 80%
- Researched & built POC on detection of probable diseases based on drug purchase patterns using K-Means clustering
- Explored patterns for drugs that were frequently bought together using Apriori algorithm, to interpret deficiency, lifestyle, nutrition
- Helped leadership plan roadmap to promote region-wise disease awareness and pharma campaigns be more specific across 3 continents
- Constructed a centralized data warehouse for streamlining data management, optimized resources by 75%
- Singlehandedly remodelled 30+ critical reports to ensure real-time data sync for 40+ trials across 3 continents, reduced latency by 94%
- Provided end-to-end optimized solution enhancing data consistency & integrity, optimised performance by reducing time of analysis by 86.67%, and saving approx. $280K annually
- Designed an automated, near-real time solution for digitizing Trial Master Files (TMF), increased efficiency by 98%
- Created dashboard & conceived metrics for TMF error rate & SLA analysis using Microsoft Power-Platform, reduced annual cost by $240K+
- Streamlined exploratory data analysis (EDA) framework across 8 error categories, eradicated need of 500+ man week of effort annually
- Drastically decreased metric creation and calculation time from 15 days to 3 hours, diminished error due to human intervention by 87%
- Exhibiting an Adverse Event Classifier model to identify potential safety issues in pharmacovigilance
- ▪ For any Clinical Trial, feedbacks received post every visit are scored based on the sentiments, using NLP and text analytics
- Using Decision-Tree Classifier to classify beneficial, adverse, serious adverse events; predicting presence of patient in next visit
Software Engineer
July 2019 - Feb 2022
Xoriant Solutions, Pune, Maharastra
- Led data-driven hospital ranking solution of a leading US-based healthcare client; optimized medicine campaign strategy by 37%
- Spearheaded a team of 4 members to develop data models with sophisticated visualisation reduced analysis time from 5 days to < 10 mins
- Integrated auto highlighted DICOM images in client’s app, improved test report comprehension, improved feedback response rate by 34%
- Implemented a near-real-time solution to detect manufacturing issue using Power-BI & computer vision, improved efficiency by 2x
- Developed auto-detection of defect areas by processing real-time camera feed, processed and stored it Azure blob storage & cosmos DB
- Created near real-time dashboard with heatmaps & trends of defects for visualisation of most affected areas using Power platform too
- Built a predictive model to identify defect prone areas for different product lines for the mattress manufacturing client using Python
- Built IoT-based real-time temperature monitoring solution for a Mexican supply chain client, deducted manual intervention by 50%
- Developed automated alerting system by processing real-time telemetry feeds from IoT sensors of transport vehicle linked to AWS IOT core
- Built pipeline to process data streams using AWS lambda & Kinesis stream & store in AWS Dynamo DB; Used Flask based APIs for retrieval
Projects/Blogs
Sometimes I start pondering upon different things that goes around me. Out of curosity, I tend to try to understand the reasons behind things. This is an endevour to document these thoughts down.
Categorization or segregation of topics in news articles
Clustering based on unsupervised machine learning approaches (K-Means, LDA)
Stock price prediction based on market sentiment
Prediction of stock prices using sentiment analysis, regression, deployed with Streamlit
Contact
Location:
Sarjapur Road, Bangalore, Karnataka
Email:
sudeshna_kundu_ampba2024s@isb.edu