Feb 2023 – Sep 2024 · 1 yr 7 mos

• Identification of Call intent: Considered various topic modeling techniques (BERT, LDA, NNF) and Facebook Llama2. Ultimately, utilized Generative AI, specifically OpenAI GPT-3.5 APIs, to identify call intents and understand the reasons for changes and cancellations of reservations with 94% accuracy.
• Web referencing: Applying Retrieval Augmented Generation (RAG) using LangChain to create a robust question-answering model based on the company's internal documents. The overarching objective is to compare two different documents on the same topic and identify similarities and any discrepancies.
• Conducted two technical workshops on large language models and prompt engineering, enhancing the skill set of the entire IT team comprising over 250 individuals.
• Orchestrated and adjudicated the team's inaugural hackathon, involving 60 students.
• Visited multiple universities to promote our department, presenting ongoing projects, and conducting interviews to assess candidates for open positions.

Jun 2022 – Aug 2022 · 2 mos · Philadelphia, Pennsylvania, United States

• Professional Network Modeling -
• Visualized the entire professional landscape in the construction domain
• Built a Lutron relationship score for each company (generalize it to quantify the relationship between any two companies)
• Developed a model to predict an Architect and Electrical Engineer that would be working on the project

Aug 2019 – Jun 2021 · 1 yr 10 mos · Gurgaon, India

• Secured first position in Digital Innovation Hackathon - Designed solution to discover spread charged by liquidity providers based on customer transactions and market data
• PnL Driver Analysis: Derived & analyzed various metrics that define bank’s profit and loss; created Dash app to provide a consolidated view at currency pair and region level with direct connection to KDB database to ensure the real-time update
• Liquidity Report: Created automated pdf report to provide an overview of how bid-ask spread, volatility & paid-given activity has trended over time & especially with pre-covid times
• Hydra Dashboard: Created a Plotly Dashboard app to showcase the trades that monetize Intra and inter-venue same pair crossing opportunities within the eFX spot space

Aug 2018 – Jul 2019 · 11 mos · Noida Area, India

• Represented EXL at TATA Crucible Hackathon'18 Finals @Bangalore - Driver Profiling and Smart Helmet
• Inadmissible Debt Prediction: Used Logistic regression to predict if a policy will turn into bad debt; estimated $280K benefits over a year for working capital gain; Accuracy: 88%
• · Sampling Analytics - Used XGBoost to assist auditors in capturing erroneous transactions through smart sampling; detecting 4 times more errors with respect to random audit

Aug 2017 – Jul 2018 · 11 mos · Noida Area, India

• Dynamic Staffing Model: Modified EXL custom client tool to generate staffing requirements as per target turn-around time, incoming volumes, exception rates, and average handling time
• Volume Forecasting Tool: Built VBA & R-based tool that automatically selects time series techniques (ARIMA, TBATS, ETS, HLT, HW) with minimum MAPE to generate volume forecast

May 2016 – Jul 2016 · 2 mos · Gurgaon, India

• • Financial Decline Rule Optimization (SAS, Excel)

Jun 2015 – Jul 2015 · 1 mo

• Developed sound understanding of marketing principles and applied them to solve cases by Harvard Business School
• The Springfield Nor'easters : Maximizing Revenues In The Minor Leagues.
• Steinway And Sons : Buying A Legend.
• Mountain Man Brewing Company : Bringing The Brand To Light.
• Analyzed a Management article "What Marketers Misunderstand About Online Reviews", published in the Harvard Business Review.