Banu Prakash Reddy Munnangi

Lead Data Scientist

Phone:

+44 7386658362

Email:

Banu.munnangi@gmail.com

Address:

London

Date of Birth:

Sep 03, 1995

A Bit About Me

Highly skilled Data Scientist with years of industry experience with hands-on expertise in applying Machine Learning and Artificial Intelligence techniques to develop and implement end-to-end solutions across industries with a primary focus in finance and insurance. Proven ability to work effectively in an agile environment, with a focus on time series modeling and classification algorithms. Strong problem-solving skills and ability to coordinate models into applications, making them ready for production with a track record in delivering results and driving business impact.

My Core Competencies

Programming Languages

Pyspark, Python, R, SQL

Cloud Platforms

Palatir - Foundry, IBM Watson, Azure ML, AWS Sagemaker, GCP

Application Frameworks

Django, Flask

Deep Learning Frameworks

PyTorch, Keras, Tensorflow

Visualisation Frameworks

Plotly-Dash, WandB, Tensorgraph

UI Expertise

Typescript, jQuery, JS

Special Traits

Natural Language Processing

Generative AI

Attention Models (Transformers - BERT, FinBERT)

A/B Testing

Webscraping - Selenium

Practical Expertise

Time series Modelling

Agent Based Modelling

Reinforcement Learning

Stochastic Modelling

Bayesian Methods -

Bayesian Optimization

EM Algorithm

Probabilistic Graphical Models

Financial Markets

Hidden Markov Models

Professional Work Experience

Dec 2020 - Present
Swiss Re, London

Significant Achievements

Guest Lecture at SCMHRD University (Dec – 2022)

Delivered a highly acclaimed guest lecture at prestigious Symbiosis Centre for Management and

Human Resource Development, on "Adoption of Data Strategies in Insurance" addressing over 250

MBA students and professors, highlighting my expertise in this field

Best Employee of the Year (Apr – 2022)

Best Mentor (Jun - 2023)

Best Team Award (Sep - 2023)

Growth Booster of C&S (Dec - 2022)

Pinnacle of Credit & Surety (C&S) (Dec - 2021)

For consistently delivering high-quality work and inspiring teams across the organization
For exhibiting outstanding performance throughout the year and significant contributions

Significant Achievements

Innovator in the NASSCOM AI GameChangers - Compendium (Aug - 2021)

For an innovative work on Drug Repurposing using AI developed at Wipro. This project has been recognized and continued by TFORD - Pune, Government of India and played a role in reinforcing the usage of "Remdesivir" against COVID-19.

Best Wipro Front Line Manager (Aug – 2020)

Awarded as a part of best First Time Manager's program for being an outstanding manager throughout the tenure delivering quality results.

The Extraordinaire Award from the CEO of Wipro (Feb - 2020)

At World Economic Forum, Davos

For excellent individual contribution in filing an IP - Revenue Prediction. This work has been heavily

appreciated at World Economic Forum, Davos - 2020, for accurate sales predictions and inferences by

Mr. Harmit Singh, CFO, Levis.

Outstanding Contribution Award at Wipro Digital Annual Meet (May - 2017, 18, 19, 20)

Received an outstanding contribution award every year that I had engaged with the organization for my consistent deliveries.

Awarded with 'Best New Employee' and 'Instant Karma' awards in Standard Bank (Dec - 2016)

For significant contributions towards Wipro's clients.

Jul 2016 - Dec 2020
Wipro, Bangalore

Lead Data Scientist (Credit Portfolio Manager)

Assistant Vice President

imgbin_swiss-re-reinsurance-finance-zurich-insurance-group-png.png

Significant Responsibilities

Founding member of the Credit & Surety squad to adopt a cloud-based analytics platform for entire data handling, analysis, reporting and data science projects.
Played a key role in setting up Quota Share, XoL reinsurance treaties and implementing them across platforms for efficient exposure relief to grow the business and has been highly appreciated for the same for saving close to 400k USD in terms of IT costs.
Automated creation of reinsurance bordereaux reports on stargate saving 2 FTE efforts.
Leading data science and analytics chapter, CorSo Analytics Network (CAN) in Bengaluru in designing the data solutions in

Corporate Solutions and acting as a consultant at SwissRe Institute enhancing the data science

chapter across the organization and exploring generative AI solutions.
Automated monitoring analysis and reports, developed production, exposure, rates, counterparty watchlist dashboards, built industry and country risk heatmaps using data from Refinitiv, that are shared with all the portfolio owners, on Palantir -Foundry.
Working on FLVs along with SRI and Andreas Hillebrand and Dr. Daniel Aunon to improve the efficiency of the ML models and to scale the models for reserving and also for optimizing the

portfolio towards efficient frontier.
Automated "Re-Pricing" activity with the help of data knowledge saving huge efforts from reserving actuaries. Collaborating with pricing actuaries in developing small portfolio costing.
Developed a portfolio monitoring tool that computes the impact of a new deal on the portfolio,

and constantly provides the movement of the portfolio direction on the risk return matrix.

Contract Bond Default Prediction

Collaborated with Prof. Dimitris Bertsimas from MIT to develop a model for predicting default on contract bonds utilizing optimal classification trees and interpretable AI.
Led the implementation and A/B testing of the model, with promising results, and worked closely with Professor Bertsimas to fine-tune and optimize the model for deployment.
Results have been significant, and the model is deployed in one of the underwriting applications after successful completion of A/B testing.

Surety Smart

Developed an NLP-based engine for surety contract analysis, which utilizes OCR to read input contracts from underwriters and allocates risk classification.
Utilized BERT model for sentence classification to extract key clauses that may cause financial or legal damages, resulting in improved risk management and cost savings for the team.
This innovative solution saved significant manual effort and helped identify potential risks in a timely and accurate manner.
Currently scaling the solution to be implemented across the entire organization.

Surety Client Search – EMEA Region

Implemented a comprehensive data analysis strategy to identify potential surety clients by combining financial information from S&P Capital IQ, news data from Factiva, and ongoing tender information from TED Europe.
Utilized financial indicators to filter clients based on their financial stability and BERT model to analyze news data to identify their surety potential.
Incorporated tender information to accurately identify hotspots or regions to focus on, resulting in an efficient and effective approach to lead generation and client acquisition.
The strategy helped to identify potential clients and increase the success rate of acquiring new clients.

Price Benchmarking

Developed a product called "Price Benchmarking," an application that assists underwriters in their deal-making process by providing them with a list of similar deals from history, based on minimal input which can be used to set an approximate price for the deal instantly
The application also provides relevant Credit Default Swaps (CDS) data and correlates internal prices with available CDS, making inferences based on current CDS using regression algorithms.

Target Liability Portfolio

Developed a "Target Liability Portfolio" that provides a strategic imperative as a part of steering activities for an industry segment by a 360-degree evaluation of the industry segments present in the portfolio which includes contribution to portfolio risk, profitability, and uncosted risk.
Utilized historical data, industry outlook, and forward-looking views on premium, exposure, loss frequency, and loss severity to evaluate the strategic imperatives, that are forecasted using Bayesian time series to provide accurate and actionable insights for the portfolio.

Automated Portfolio Deal Monitor

Built a special portfolio deal monitoring platform which consists of automated obligor fuzzy matching, existing limit and consumption extraction from Group CRM for the list of obligors, performing strict customised checks against the eligibility criteria of the portfolio.
This automated extraction helped the team to perform portfolio costing in quick fashion making a way for increased portfolio deals.
Built a dashboard showing the portfolio evolution and compositions and information regarding recasting of portfolios on a quarterly basis.

Data Science Manager

Team Lead - Advanced AI Practice

Drug Repurposing using AI

Led and mentored a team of 8 data scientists in designing and developing an advanced algorithm for drug repurposing using AI for COVID-19 while working inline with Prof. Russ B. Altman's project from Stanford University.
Utilized 3D modeling to predict protein-protein interactions and regulate them from drug target interactions. The drugs identified were validated against WHO solidarity trials.
Won NASSCOM's AI game changer award under innovator category for this and an IP was filed.

Wipro P & L prediction - Revenue and Cost Prediction, FP&A

Held complete ownership of an aristocratic project at Wipro, where revenue, costs, order bookings, and profit & loss of the organization were predicted for the succeeding quarters with an accuracy ranging between 98% - 99.9%. This project involved understanding of end-to-end business flow in the organization and was directly handled by CFO and FP&A team.
Initially started with Revenue Prediction and its success and accuracy led to cost prediction and order booking prediction, and finally aimed at automating guidance sheet in financial report.
Involved work with Multivariate Time Series, LSTM, AutoML using Keras and Hyperas, Bayesian
Optimization, Financial Stochastic Calculus and Cyclic Learning Rate, and the organization was able to file the 'IP' on the same.
Provided Causal Inferences at multilevel granularity that helped CFO team to get more insights in revenue and cost flow especially in case of group customers and non-manpower costs.
Developed NLP search engine for the same application addressing search queries by converting most of the natural language queries to SQL queries using Rasa NLU with Spacy and duckling configuration.
Written Restful APIs in Flask and used Oracle DB, MySQL for data storage and access.
This project helped the organization to win clients like 'Levis' for their sales prediction.

Levis Sales Prediction

Developed time series models that Predicted Levis Men’s bottom sales over US and Europe region in both ecommerce and brick and mortar channel with an accuracy greater than 95% which included net sales, gross sales, and Dilution(discount). Also, provided the client with optimal dilution based on casual inference.
This work has been presented at World Economic Forum in Davos, 2020 by CFO of Levis.
Mentored developers of the project so that they can handle the future requirements of the client.

Feature Extraction, Albea

Extracted the features affecting the decline in the production of lipstick containers and the factors that led to scrap in the production by using the concepts of autoencoders.

Automation - Django Application Development, Standard Bank

Presented a white paper on “Multiple Django database with backup routing accessibility”.
Part in developing bank’s ‘User Access Provisioning’ application that automates a huge amount of
manual work in creating user branches in SAP and Mainframe using Python Django.
Developed a standalone executable application for remote workstation access using Tkinter

Banu Prakash Reddy Munnangi

A Bit About Me

My Core Competencies