I am a highly-skilled Machine Learning Engineer with over 3 years of experience in applying data science and machine learning algorithms to solve complex business challenges. My expertise lies in Natural Language Processing (NLP), Deep Learning, Generative AI, LLMs, and cloud computing technologies such as AWS and GCP.
In my current role at Liaison Medicare Pharma, I have successfully implemented a salesperson pitch audit mechanism using BERT Embeddings, Semantic Search, and Logistic Regression. This system achieved an accuracy of 93% in analyzing sales pitches and conversations during meetings. I have also developed an NLP system for context data extraction from sales conversations, reducing LLMs token cost by around 78%. Additionally, I have worked on sales forecasting time series analysis using LSTM and have experience in developing Flask-based REST APIs and Microservices architecture using AWS services such as SQS, SES, Lambda, S3, Sagemaker, and Vertex AI.
Prior to this, I worked at Aon as an Analyst, where I designed a machine learning model to predict the estimated risk score for proposed insured customers. I also oversaw troubleshooting and resolution of data configuration issues in ETL pipelines and collaborated with the Business Intelligence (BI) team to provide technical support to internal teams and stakeholders. Furthermore, I initiated the implementation of a Langchain-based layer for multiple LLM and Generative AI use cases.
As an Associate Data Scientist at Syntax Edutek, I engineered a system for live monitoring of lecture classes using YOLO, OCR, and Realtime Topic Detection, saving significant man-hours of class auditing. I also improved a coupon recommendation system using content filtering, Collaborative filtering, and Multi-Armed Bandit Reinforcement Learning algorithm, resulting in a 20% increase in coupon usage. Additionally, I performed sentiment analysis of course reviews using Hugging-face, reducing manual review reading by 70%. I have experience in developing ML models for various use cases using machine learning algorithms like Regression, Naïve Bayes, KNN, Clustering, Deep Neural Networks, CNN, and Random Forest. I have also optimized models using techniques such as pruning and quantization, making them 80% lighter and able to run on edge devices.
