Vyshnavi Challapalli

About Vyshnavi Challapalli

I’m a Machine Learning & AI Engineer with 4 years of experience building scalable, production-grade ML and GenAI systems across NLP, computer vision, and time-series use cases.I’ve deployed LLMs (BERT, RoBERTa, BART, GPT) using LoRA/QLoRA, RAG, and Hugging Face, processing 1M+ documents and 500K+ daily inferences in cloud environments.My experience spans end-to-end ML systems from distributed data pipelines (Spark, Kafka, Databricks) to MLOps and deployment using Docker, Kubernetes, MLflow, CI/CD, and FastAPI on AWS, Azure, and GCP.I enjoy working on problems where model performance, scalability, and reliability matter, and I’m especially interested in LLM/GenAI, applied ML, and production engineering roles.

Experience

1. Ai Engineer

   HCLTech

   Aug 2025 — Present · US

   Developed transformer-based LLMs (BERT, RoBERTa, GPT) using PyTorch, Hugging Face Transformers, PEFT (LoRA, QLoRA), RAG with Pinecone, ChromaDB vector databases, achieving 92% accuracy in NLP tasks on 1M+ documents via AWS SageMaker, Lambda.Led IT project for enterprise document automation using BART, GPT fine-tuning with LoRA on AWS S3, EC2, integrating OCR, NER via Hugging Face, PyTorch, automating 500K+ docs daily into PostgreSQL with 98% extraction accuracy.Delivered cloud migration IT initiative migrating legacy ML modelsto Azure ML Studio, GCP BigQuery using Docker, Kubernetes, MLflow versioning, PySpark ETL, cutting infrastructure costs by 55% for 10+ US client applications.Engineered end-to-end MLOps pipelines with MLflow, Kubeflow, Docker, Jenkins CI/CD, Git for model versioning, automating supervised, unsupervised learning (XGBoost, Random Forest) workflows on AWS EC2, S3, Azure ML Studio, reducing latency by 35%.Implemented anomaly detection, recommender systems using Scikit-learn (Isolation Forest, SVM), LSTM, RNN neural networks in TensorFlow, Keras, hyperparameter tuning with Optuna, integrated with FastAPI for real-time inference on Kubernetes.Designed ETL, ELT pipelines with Apache Airflow, PySpark on Databricks, Kafka streaming, Hadoop for big data processing into MySQL,PostgreSQL, MongoDB, enabling EDA with Pandas, NumPy, visualization in Power BI.Built fine-tuned LLMs with prompt engineering, semantic analysis, NER, embeddings, tokenization via Hugging Face, deployed on GCP AI Platform, BigQuery with A/B testing, ensembling (Gradient Boosting, LightGBM), cutting bias by 40% per SHAP analysis.Formulated reinforcement learning agents, GANs using PyTorch, TensorFlow, feature engineering, selection in R, SQL, Bash scripts,orchestrated via Apache Spark on AWS, GCP, predicting disruptions 48-72 hours ahead at 85% accuracy.

Education

• Saint Francis University

  Master's degree, Information Technology