Hire Pradeep V. - Fine-Tuning Engineer

Designed and deployed AI/ML solutions using Vertex AI, Cloud Functions, Cloud Run, Dataflow, and BigQuery to optimize healthcare claims processing, automate document understanding, and enable intelligent clinical data retrieval. Built machine learning and deep learning models using TensorFlow, PyTorch, CNNs, RNNs, Transformers, and ensemble methods like XGBoost. Applied NLP techniques using SpaCy, NLTK, BERT, GPT, VADER, TF-IDF, and topic modeling (LDA, NMF) to extract insights from unstructured healthcare text, enabling sentiment classification and automated routing. Fine-tuned LLaMA and Mistral models using LoRA and QLoRA, improving healthcare document understanding and reducing model latency. Developed Generative AI systems with Vertex AI, Hugging Face integrations, and Agentic AI workflows using LangChain, LangGraph, CrewAI, and ReAct reasoning patterns. Implemented semantic search using vector databases (FAISS, Pinecone, ChromaDB) and architected Retrieval-Augmented Generation (RAG) systems supporting millions of healthcare documents. Built serverless ETL pipelines using Dataflow, BigQuery, Cloud Functions, and Airflow for training and inference workflows. Worked with relational and NoSQL databases including PostgreSQL, MySQL, MongoDB, Cassandra, DynamoDB, and ElasticSearch. Established MLOps workflows using MLflow, Docker, Kubernetes, Vertex AI Pipelines, CI/CD, and Cloud Monitoring for scalable deployment and tracking. Implemented security, IAM configurations, VPC isolation, and responsible AI techniques including guardrails (LLMGuard, TruLens). Performed A/B testing, cross-validation, model monitoring, and Champion/Challenger evaluations for production models. Collaborated with engineering, business, and product teams in Agile sprints, code reviews, and stakeholder demos.

* Built and deployed ML models for regression, classification, and clustering on structured and unstructured datasets. * Applied supervised and unsupervised learning techniques including XGBoost for predictive analytics and pattern extraction. * Engineered features and optimized model performance using GridSearchCV and Optuna. * Developed deep learning solutions using CNNs, LSTMs, and GANs for computer vision and NLP tasks. * Fine-tuned BERT and T5 models using Hugging Face Transformers for financial document classification and entity extraction, achieving high F1-scores across 50+ document types. * Implemented prompt engineering techniques for GPT-3.5 and GPT-4 to automate report generation and reduce manual review time. * Performed data preprocessing using Pandas and NumPy to ensure high-quality inputs for modeling. * Built ETL pipelines using Databricks, PySpark, Azure Data Factory, and Azure Functions. * Deployed ML models using Azure ML, Cognitive Services, Synapse Analytics, and containerized them using Docker and Kubernetes. * Implemented CI/CD, model monitoring, and automated retraining processes using Airflow and MLOps practices. * Built Tableau and Plotly dashboards to visualize model performance and insights

* Developed data pipelines in Python using Pandas and NumPy for data cleaning, transformation, and preprocessing. * Built ML models using scikit-learn including Linear Regression and Logistic Regression. * Performed large-scale data processing using Spark and Hadoop. * Wrote complex SQL queries to extract, join, and aggregate data for analysis. * Built text classification models using TF-IDF and fine-tuned BERT models for NLP tasks on customer service transcripts. * Applied VADER sentiment analysis to identify sentiment trends and prioritize responses. * Designed Tableau dashboards for KPI tracking and business reporting. * Conducted EDA, descriptive statistics, and K-means clustering to segment customer datasets. * Built APIs and conducted web scraping using BeautifulSoup. * Followed Agile practices using Jira and Git for task tracking and version control.