Akarsh Singh Email & Phone Number

Boston, Massachusetts, United States

Boston, Massachusetts, United States

Generated NLP summaries and performed diagnostic analytics (including frequent sequential patterns and association rules mining) on EHR Data of 5504 cardiovascular disease patients to predict amputation-likelihood and suggest digressive treatments/diagnosis to intercept prevent life-threatening surgeriesEnhanced state-of-the-art deep learning models for grading prostate cancer using whole slide images of core-needle H&E stained biopsies using northeastern high performance computing clusters. Current model achieved an agreement of 0.935 (quadratically weighted kappa) and 0.862 (qwκ, 95% confidence interval (CI), 0.840–0.884) on internal and external test sets.Developed and deployed a dynamic, interactive web-application using Flask (Python) for research inference and data collection. Ultimate goal to significantly reduce prostate cancer grading time for pathologists but providing them grade predictions (with state-of-the-art accuracy), visualizations of biopsies with highlighted areas for primary inspection, summarized patient history using NLP, etc.

Boston, Massachusetts, United States

Creating ML/Agent-Based Models to predict COVID-19 spread using human mobility patterns. Wrote optimized SQL queries to extract precise information (ETL) from over 4+ Billion Rows and 18 Tables.Securely streamed data from external AWS S3 Bucket and reduced overall data collection runtime by 25%.Contributed to analysis and research reports to help the lab secure future grants and funding for the project.

Framingham, Massachusetts, United States

Wrote efficient SQL queries to build 30+ dashboards for monitoring KPIs at 18 Staples fulfilment centers.Migrated over 3800 SKUs to robotic picking stations by building a custom ETL pipeline and an automated Python script. Statistical analysis and clustering methods with various distance measures resulted in a 78% transfer success. Replaced traditional rule-based release logic (order fulfillment) in 4 subsections via data-driven methods. Two Decision trees trained to enhance SKU affinity and carton-station assignment process reduced cut-times by 11%.Traveled to 2 Fulfillment Centers to exchange feedback from cross-functional on-site teams to improve reports, algorithms, and human engineering processes.

Mumbai, Maharashtra, India

Tutor Creed is an Ed-Tech startup with a goal to subsidize expensive after school educationforIGCSE/IB board students by streamlining online portals, educational content & online/offline tutoring.Communicated innovative data-driven solutions by identifying KPI’s from visual dashboards (Tableau) with student data for operations, business, marketing and pitched business cases to 7 executives in 75% of proposed projects. Deployed mix marketing models for customer churn and acquisition leading to 30% YoY revenue growth. Effectively distributed funds of over $15000 in value for offline setup and marketing based on data driven market research and GA4 data.

Remote

Worked intently in research and development (Sports Analytics) to aid in the deployment of a fantasy team prediction model. Devised end-to-end agile deployment of reliable ML models, LSTM-RNN based cricket-match simulation and ARIMA to estimate player fantasy points.Optimized final fantasy teams based on multivariate optimization using genetic algorithms on Python (NSGA-2) implemented from scratch without using librariesAssisted in the deployment of models using AWS Lambda and Postman for API development.The final API model deployed had an accuracy of over 64% for fantasy team predictions.

Mumbai, Maharashtra, India

Created and successfully deployed a translator tool for an international client capable of parsing foreign language pdf files (German, Spanish, German, etc.) and translating the contents into English by creating a new file with a txt extension using NLP techniques and APIs. Created code that extracted essential information from invoices and stored them into a database rejecting duplicate entries (SQL or Excel). Initiated by label-based extraction using regex, tabula, and pandas on python.Created a tool for client usage, translated and extracted data from over 1000 invoices, and reported reasons for specific discrepancies in this data.

Master Of Science - Ms, Data Analytics Engineering, 3.9

Minor Degree, Data Science/Analytics Specialisation, 22 Extra University Credits (90%)

Relevant Courses- Data Analytics for Business, Analysis of Algorithms, Machine Learning: Regression, Classification, Clustering, DBMS

Bachelor Of Technology - Btech, Mechatronics And Robotics, 3.2/4

Activities and Societies: Student Council Committee Member (2017-18)Relevant Courses- Engineering Mathematics (I, II, III and IV).