Specialist in language modeling, NLP, large language models (LLMs) etcManaged multiple model building software projects for various business problems of a major e-commerce businessTechnical direction, design, and implementation of predictive machine learning math models to support many business functions:• Built deep learning models for Natural language processing (NLP) using BERT (transformer) and large language models (LLMs) like GPT-3, T5, T5-FLANo Modeling for improving semantic search relevance and product recommendations using cutting edge NLP techniqueso Programming OpenAI and OLlama LLMs for evaluating and improving the relevance of recommended furniture products .o Improving the relevance of search queries typed by consumer on website for furniture productso LLM prompt programming and tuning, ColBERT, DSPy, Haystack, FAISS o Natural language processing to analyze texts using BERT (transformer) and instruction-tuned models like GPT, T5-FLAN.o Text generation using large language models like GPT, Bloom, T5-FLAN and feature attribution for Roberta text predictions.. T5-FLAN model tuning for various NLP uses. Reduced need for text data for training by over 1000 timeso NLP modeling for sentiment recognition in textual surveys and comments using Stanford Sentiment Treebank (SST)• Modeling for growth marketing: Built predictive models to predict user acquisition / lead generation.• Modeling for patient outcomes prediction using data from biomarkers and phenotypes using classical and DNN techniques in tensorflow• Models using techniques from reinforcement learning with applications to advertising technology. Web consumer focused modeling for product recommendations, semantic search and website personalization.

Designed and implemented predictive machine learning models and data analytics to solve a variety of business problems in corporate finance, manufacturing yield improvement, statistical process control, and financial forecasting. Business impact of my models was savings to GE of over $300 millionAwarded: Technology Titan 2019 Digital Engineering Excellence 2019Software engineering and hands-on project management for machine learning solutions and deep learning models. Programmed machine learning modeling projects in pytorch, tensorflow, R, and Matlab for:• corporate finance. - using statistical machine learning techniques• manufacturing process yield improvement - using traditional machine learning techniques• computer vision algorithms based image classification to identify bad jet engine parts - using deep learning techniques of CNN, generative models• forecasting for multi-variate time series. - using deep learning techniquesLed and managed globally distributed teams of talented data scientists to execute projects

Researched and designed machine learning models for e-commerce coupon recommendation system to increase revenue from website visitors, to predict customer affinity for brands and products and to increase website revenue by serving more relevant ranked brands /product item sets in real time. Text mining / natural language processing, deep learning with word embeddings and incorporating time series of sales quantity history. Filed patent number 14921068 in October 2015. Architected and implemented the system for high performance in Apache Spark.

Managed software development, data analysis and math model implementations for forecasting economic and Basel regulatory capital, leading a global team of quantitative analysts / software engineers• Partnered with executives, business stakeholders and business analysts (product managers) to define roadmaps for porting archaic SQL-based risk models for market and credit risk to a modern infrastructure using R, python, and Hadoop• Quantitative analysis and programming in R/python of econometric math models

Managed software engineering projects in Java for this leading fixed-income asset manager. Thought leadership and insight communication between bond portfolio managers and software engineering team, defining requirements and functional design specs, and managing software developers and testers. Worked with traders and portfolio managers, defining requirements for front-office portfolio management tools, translating them into functional design specs.

Directed a group of quantitative analysts in the development of quantitative/math models for internal software analytics for OTC Derivatives portfolios.