Pavan Kumar M

AI/ML Achievements at ADIB Bank

Autonomous AI Systems

• Led development of multi-agent RAG document intelligence system (40% faster compliance reviews, 25%↑ accuracy) using Llama-2, LangChain & PG Vector
• Created Agentic RAG wealth advisor generating hyper-personalized investment advice (60% workflow reduction, 35%↑ engagement) with OpenAI fine-tuning + RLHF
• Developed voice-enabled Agentic RAG system integrating transaction logs, policy docs, and MS Accelerator-powered voice-to-text, achieving 90%+ retrieval accuracy with real-time sentiment analysis and document summarization

ML Optimization

• ATM cash forecasting: Hybrid LSTM+Prophet model reduced costs by 18%
• Churn prediction: XGBoost+SHAP model cut attrition by 22%

• Developed an ML model (XGBoost/LSTM) to predict last-minute special assistance requests, reducing gate delays by 15% by analyzing historical booking patterns, flight connections, and passenger demographics.
• Skywards Customer Tier Prediction Engine | Developed an ML model (XGBoost) analyzing 50+ behavioral features to predict member upgrades/downgrades with 88% accuracy, enabling targeted retention campaigns that reduced attrition by 22% through personalized interventions.
• As part of Supply chain Management, forecast Meal Up-Lift per aircraft. And perform Text Analytics and inventory Planning.
• As part of POC, Predict categories of Aircraft Engineering parts based on their images Consisting of 80 diverse classes.
• Developed scalable classifiers and tools by deploying Machine Learning, data regression and rule-based models.
• Deployed Predictive Modelling techniques including ML & advanced statistics to conceptualize scalable business solutions.
• Developed and deployed a state of the art 4 layer neural network.
• Processed terabytes of data via Machine Learning & Predictive Modelling to develop dynamic solutions.
• Successfully trained the neural network with Computer Vision towards Image Classification & Object detection.
• Developed state-of-the-art large-scale ML Services & Applications on the Cloud by applying predictive technology.
• Processed structured/unstructured data & gained insights by leveraging predictive modelling, statistics, ML & data mining.
• Independently executed project lifecycle management to implement scalable ML solutions & document results.
• Key Achievements: AED 5M in Meals uplift (Aircraft) optimization revenue in 5 months with Business class travels. Improved 10 times more service rate for Special handling PAX at Emirates Airlines.

• Analyzing public transportation data (Taxi, Bus, Train) to measure and identifying Accuracy, Loading and headway, route density finding insights with spatial data analysis and passenger demand forecasting.
• Deployed ML algorithms like k-NN, Naive Bayes, SVM, Random/Decision Forests, Logistic Regression etc.
• Coordinated with Product teams to implement algorithms for enabling users and creating developer-facing products.
• Employed Agile methodologies & adopted standard ML techniques for exploiting modern parallel environments.
• Deployed the latest DL algorithms to user modeling, topic modeling and engagement prediction problems including Regression, classification, learning to rank, Transferred Leaning, etc.
• Developed deep learning network architectures & DL-based models to evaluate and assess model accuracy.
• Designed prototypes by manipulating and analyzing complex & high-volume of data from various sources.
• Coordinated with the Engineering team for processing huge datasets & incorporating the models on the platform.
• Deployed mathematical, statistical and programming expertise to formulate strategy using a deep learning framework.

• Customer behavior analysis and sentiment analysis for banking products and policy, and then look for specific solutions in the Data Science domain.
• Also worked on Banking Infra structure asset data of SCOM for correlation between performance, alerts and events data and Automated Analysis - The shout cut to proactive monitoring.
• Led the development of Machine Learning algorithms & enhanced accuracy of existing algorithms.
• End-to-end project execution from Requirement gathering, project development to deployment, and project management.

• Worked as analyst for two clients for analyzing Customer catalog information and Telecom information which is spatial data.
• Led the development of Machine Learning algorithms & enhanced accuracy of existing algorithms.
• Broadened the scope of existing algorithms by deploying various Oracle products.
• End-to-end project execution from Requirement gathering, project development to deployment, and project management.
• Boosted the run-time of existing algorithms & developed the necessary infrastructure to operate & maintain algorithms.

• Analyzing data of job portal and extracting descriptive statistics for active and inactive users and visualize the user behavior and analyzing Health asset data for optimizing the cost.
• Involved in the development of exploratory analysis using different statistical tests.
• Visualize the data using Tools and process the data using data-pipelines.