Home

Yashwanth Reddy Vasireddy - AI/ML engineer

[email protected]

Location: Remote, Remote, USA

Relocation: Yes, anywhere in mainland US.

Visa: OPT

Resume file: Yashwanth_Reddy_AI_1770151390237.pdf Please check the file(s) for viruses. Files are checked manually and then made available for download.

Summary: AI/ML Engineer and Data Scientist with 3+ years of experience building production ML systems for complex enterprise environments. Specialized in developing NLP pipelines and fine-tuning Transformer models to handle sensitive, large-scale unstructured data. Deeply skilled in deploying RAG architectures and optimizing inference latency using Python, PySpark, and AWS. Proven track record of translating messy real-world data into scalable solutions that automate workflows and drive operational efficiency. Skills: Languages: Python, R, SQL Libraries & Models: TensorFlow, PyTorch, Keras, Scikit-learn, Pandas, NumPy, Seaborn, Matplotlib, Beautiful Soup, Hugging Face, Lang Chain, Lang Graph, NLTK, OpenCV, Spacy, ScispaCY, Transformers (BERT, GPT, RoBERT), Crew AI Databases: MySQL, PostgreSQL, MongoDB, ChromaDB Cloud Services: AWS, IBM Cloud Services(Auto AI, Watson Assistant), Azure, GCP (Big Query) Artificial Intelligence: Supervised & Unsupervised Learning, Reinforcement Learning, Deep learning Architectures (Neural networks, Transfer Learning), NLP (Semantic Analysis, Text segmentation), ML Ops (ML Flow), LLMs Data Engineering: PySpark, Airflow, Databricks, ETL/ELT Pipelines, Data Modeling, Hadoop, Teradata, Snowflake Tools: DataBricks, Tableau, Power BI, Spark, Docker, Google Colab, GitHub Others: SDLC, Agile, Jira, CI/CD Pipelines Experience: AI Engineer | EY, USA | Jan 2025 Present Designed and deployed Python based AI model to automatically detect and classify sensitive data (PII, PHI, financial records) across emails, documents, and unstructured text, helping reduce unintended data exposure by 45%. Built NLP driven classification pipelines using transformer modelsto understand content context, significantly reducing false positive DLP alerts by 30% and improving trust in automated security controls. Implemented supervised and hybrid machine learning modelsto analyze large scale data movement patterns and identify abnormal data flows linked to potential leakage incidents. Engineered contextual features using custom tokenizers and embeddings, handling noisy unstructured text from emails to improve classification accuracy on sensitive documents. Finetuned BERT based transformer models for sensitive data identification, improving classification precision and recall while reducing inference latency by 25% for near realtime security scanning. Integrated the AI powered DLP classifier with security dashboards and alerting workflows, enabling security teamsto monitorrisk trends, investigate alerts faster, and respond efficiently to potential data exfiltration events. Data Scientist | HCL, India | Feb 2021 Jul 2023 Built and deployed predictive ML models (Random Forest, Logistic Regression, KNN, K-Means, PCA) to identify fraud risk and customer behavior patterns, reducing fraudulent claims by ~25%. Automated SQL-based data pipelines to refresh daily customer profiles, policy eligibility, and quote history datasets, enabling near real- time model scoring and analytics. Designed and analyzed A/B experiments comparing rule-based vs. AI-driven recommendations, resulting in a 12 15% increase in offer acceptance rates. Built executive-facing Tableau dashboardsto track model performance, fraud trends, and recommendation impact, improving leadership visibility and data-driven decisions. Developed recommendation models using Python and TensorFlow to personalize insurance product suggestions, improving relevance and decision accuracy. Applied NLP techniques to analyze customer service transcripts and sentiment trends, contributing to a 10% uplift in customer satisfaction scores. Conducted exploratory data analysis to uncover seasonality, claim surges, lapse risks, and fraud indicators, directly influencing personalization and risk models. Performed model validation, feature selection, and hyperparameter tuning to ensure stable performance, regulatory compliance, and reliable production outcomes. Education: Master of Science in Data Science | Rowan University, Glassboro, NJ, USA May 2025 Bachelor of Technology in Artificial Intelligence | Vidya Jyothi Institute of Technology, Hyderabad, India May 2023 Projects: Intelligent RAG & Agentic Q&A System | Stack: LangChain, HuggingFace, ChromaDB, Python, Vector Databases Built an LLM RAG system using ChromaDB and flan-t5, achieving high precision (>95%) in toolselection for complex queries. Optimized vector embeddings and text splitting and implemented strict context-windowing to minimize hallucinations and ensure fact- based retrieval. Retrieval-Augmented Generation (RAG) Pipeline | Stack: Python, ETL Pipelines, Data Modeling, Embedding Models, Open-Source LLMs Built an end-to-end RAG pipeline using multiple embedding models and open-source LLMsfor contextual query answering. Evaluated performance on a custom QA dataset and analyzed hallucination cases. Compared RAG vs. non-RAG outputs to measure factual accuracy and retrieval effectiveness. Keywords: continuous integration continuous deployment quality analyst artificial intelligence machine learning business intelligence rlang New Jersey