Sugumar S Email & Phone Number

Chennai, Tn, In

Chennai, Tamil Nadu, India

Machine Learning Engineer/Data Scientist: Design, build, and deploy ML models to analyze complex datasets. Perform data preprocessing, feature engineering, model evaluation, and optimization. Collaborate with teams to implement predictive solutions and provide insights for data-driven business decisions.Currently focusing on Natural Language Processing (NLP) and Large Language Models (LLMs), specializing in the development and fine-tuning of custom models for various text-based tasks. Working with advanced architectures like BERT, RoBERTa, Gemini, PaLM, and LLaMA. This includes tasks like emotion detection, text generation, and formal/informal text analysis, with an emphasis on improving model accuracy and evaluating using metrics like precision, ROC, and BLEU scores.

• Business Problem: Tackled volatile pricing in the raw spice material market.• Solution: Implemented a time series forecasting solution using ARIMA and Prophet models.• Outcome: Improved forecast accuracy by 90%, leading to a 10% reduction in production costs and 20% savings in procurement and inventory management.•Project Flow:Project: Raw Metals and Minerals Price ForecastingForecasted prices for eight metals using 348 records spanning 3-4 years. Conducted grouping, EDA, missing value checks, trend imputation, and outlier detection using IQR. Analyzed seasonality and trends. Trained ten models per metal, evaluated with MAPE, and deployed SVR on Streamlit for local access.

• Business Problem: Analyzed emotional tone in job candidates' communications.• Solution: Built an NLP model using BERT and RoBERTa for emotion detection and language analysis.• Outcome: Achieved 90% accuracy, enhancing emotion recognition and reducing decision-making time by 15%.•Project flow:Analyzed emotions, grammar, filler words, and formality using Kaggle and GitHub datasets. Preprocessed data with regex and lowercased text. Tokenized data for custom models, trained with BERT, RoBERTa, and GPT-2 for 16 emotions. Utilized Gemini, Palm, and LLaMA for LLMs. Evaluated with accuracy, precision, ROC, and BLEU scores. Deployed using Streamlit.

Professional Programme On Data Science And Ai, Artificial Intelligence

Currently working on training custom models using BERT, RoBERTa, and other Large Language Models (LLMs). Focusing on fine-tuning models.

Bachelor'S Degree, Computer Applications, 8.3

Bachelor of Computer Applications (BCA) from Kalasalingam Academy of Research and Education provides a comprehensive curriculum to develop.