Resume
This section contains details about my professional experience (click here for details about my personal projects). If you want to download a concise version of my resume please Click Here.
SUMMARY
Hi! My name is Ashish. I am a Data Scientist with 10+ years of experience in providing insightful data-driven solutions to customers in multiple domains such as energy, media, networking, healthcare and construction. These end-to-end solutions include data mining, ETL tasks, data exploration/visualization, predictive modelling (using statistical methods and machine learning algorithms) and model maintenance/retraining. I also love to share my knowledge through blogs and personal projects.
A few of my key project includes:
- Developed and deployed the world’s first combustion tuning machine learning model (Digital Twin) of Mitsubishi Hitachi Power System’s (MHPS) Ultra Super Critical Coal Fired Boiler.
- Developed a python based application for filtering and extraction of data from online content, resulting in the reduction of manual efforts by 80%.
- Development of multiple chatbots using RASA and Microsoft Bot Framework and its deployment on Microsoft Azure.
EMPLOYMENT TIMELINES
KEY PROJECTS
World’s First Combustion Tuning Machine Learning Model of 800 MW Coal Fired Utility Boiler
Develop an ML-based Digital Twin of 800MW coal-fired utility boiler to reduce time and cost for Combustion Tuning, prescribe input settings to meet emission norms for flue gases and get real-time prediction for key parameters.
- Performed ETL task on data from ~1500 sensors.
- Performed data cleaning activities such as handling missing values, outlier analysis and feature engineering.
- Performed Exploratory Data Analysis to generate insights from data and validated it with domain experts.
- Created data-driven models using R and Python for ~30 KPIs with real-time prediction.
- Optimized settings for efficiency during combustion tuning.
- Handled a team of UI developers, integrated complete system and deployed solution in the real environment.
Reduced Combustion tuning time and cost by 80% and deployed the solution at 2 of the power plant in Japan and Taiwan.
Email Relevancy Application
Develop a python based application for filtering and extraction of data from online content (60 shared mailboxes) and deploy the solution on Microsoft Azure.
- Extracted emails from 60 shared mailboxes using exchangelib and applied business rules to check their relevancy.
- Analyzed email body and subject to reduce Azure translation cost by 98% with an accuracy of 99.7%.
- Created Power BI dashboard to monitor application performance.
- Developed multiple logic apps to automate manual tasks.
This application reduced manual efforts of classifying emails by 85%.
Predict promotional discount applicability on a deal
Predict if a promotional discount will be added to the deal for a networking company.
- Perform ETL task on customer sales data for 1 year.
- Identified promotional discount trends across regions.
- Predicted promotional discount percentage for each deal.
Predicted promotional discount percentage with an accuracy of 97%.
Inventory Management and Sales Forecast for Alcoholic Beverages
Create a case study for Inventory management and sales forecasting of alcoholic beverages across a US state.
- Used a public dataset and classified beverages into 4 major categories
- Identify inventory trends of each category to get an idea about inventory management of a new store
- Forecasted sales of these beverages for a dealer.
Achieved an accuracy of 91%.
Duplicate Report Identification
Identify duplicate and similar reports to reduce migration efforts for a payroll service company.
- Extracted reports metadata from XML format.
- Used cosine similarity and a custom algorithm to identify duplicate and similar reports.
Reduced report migration efforts by 20%.
Exploratory Data Analysis (EDA) to identify root cause of dialysis failure
Identification of root cause for failure of dialysis, for a dialysis instrument manufacturing company.
- ETL operations on data of 100 dialysis sessions from 10 different machines.
- Visualize trend across failed dialysis sessions and identified root cause.
Received appreciation for identification of root cause and visualizations used to explain the findings.
Credit Insurance Bot
Develop a chatbot to answer queries related to credit insurance to reduce load on customer care.
- Created a chatbot using Microsoft Cognitive Services such as LUIS and QnA Maker.
- Deployed the solution using Microsoft Bot Framework and Bot Composer.
- Provided training to customer for future enhancement of corpus and addition of new intent.
This bot is successfully deployed on customer’s website.
HONORS AND AWARDS
| Bravo - Individual Awards [Oct 2021] | Excellent work ethics, delivery beyond expectations and availability round the clock. |
| Bravo - Individual Awards [Jun 2021] | Development and deployment of Chatbot for a customer in Credit Insurance domain and managing the customer independently. |
| High Five - Individual Awards [Aug 2020] | Excellent work and presentation on Chatbot. |
| Best Team Award [Jul 2018] | Successful delivery of project and achieving Customer Satisfaction Index (CSI) as 100%. |
| Certificate of Commendation [Nov 2017] | Efforts towards successful development and implementation of World's First Combustion Tuning Machine Learning Model of Mitsubishi Hitachi Power System's (MHPS) Ultra Super Critical Coal Fired Boiler. |
EDUCATION
Dr. A.P.J. Abdul Kalam Technical University
Electrical Engineering
Lucknow, India
LANGUAGES
DIGITAL / INTERPERSONAL SKILLS
| Cloud Computing | Microsoft Azure | Amazon Web Services (AWS) (Basic) |
| Programming Language | Python | R | VBA for Excel |
| Database | Microsoft SQL Server | PostgreSQL |
| Statistics | Predictive Analytics | Inferential Statistics | Descriptive Analytics |
| Modelling Technique | Statistical Modelling | Machine Learning | Deep Learning |
| Machine Learning | Supervised Learning | Unsupervised Learning | Semi-Supervised Learning |
| Algorithms | Decision Trees | Random Forest | XgBoost | Linear Regression | Multivariate Adaptive Regression Splines (MARS) | Partial Least Squares (PLS) |K-Nearest Neighbors (KNN) |Logistic Regression | Support Vector Machines (SVM) | SARIMAX | Artificial Neural Network (ANN) (Basic) | Convolutional Neural Network (CNN) |
| Unsupervised Learning | K-Means Clustering | Hierarchical Clustering | DBSCAN | Principal Component Analysis |
| Reporting Tools | Power BI |
| NLP | ChatBot | Rasa | Text Classification | Sentiment Analysis |
| Data Science Libraries | Pandas | Numpy | Matplotlib | Seaborn | Scikit-learn | NLTK | PyTorch | Fastai | Vader | Tkinter (GUI) | pyPDF2 | Dask |
| Web Development | HTML | CSS |
| Operating System | Windows | Linux |
| Interpersonal Skills | Leadership | Willingness to Learn | Collaborative | Enthusiastic | Open Minded | Self- confident | Positive Attitude | Good Listener | Open for Feedback | Analytical Mindset | Problem Solver |
DOWNLOAD RESUME
Click Here to download concise version of the above resume in the below format.