N
Nuthan Kishore Maddineni
About
Detail
Gujarat, India
Résumé
Jobs
verified_user
0% verified
Data ScientistDatabricksApr 2025 - Current (1 year 5 months)• Designed and deployed an end-to-end hybrid recommendation engine on Databricks Lakehouse, integrating collaborative filtering, content-based models, and Generative AI (LLM APIs) to boost user conversions by 15%. • Engineered scalable ML pipelines for data ingestion, preprocessing, and transformation using PySpark, Delta Lake, and Databricks Workflows, ensuring high-quality datasets and accelerating time-to-market by 30%. • Applied advanced ML techniques such as K-Means clustering, regression models, and AutoML hyperparameter tuning via MLflow, achieving 87% accuracy in predicting user preferences. • Implemented batch and real-time inference pipelines with Databricks Jobs, SQL Warehouses, and MLflow model registry, incorporating model- TLLMOps QA EngineeringTutorify AINov 2024 - Mar 2025 (5 months)• Validated a production-grade LLMOps pipeline using Python, SQL, and LangChain to ensure data quality and reliable preprocessing of structured/unstructured data from educational platforms. • Automated ETL workflows with Apache Airflow, embedding data quality checks, schema validation, and anomaly detection before ingestion into vector databases (FAISS, Pinecone). • Conducted QA testing and benchmarking of LLM workflows, fine-tuned models with Hugging Face and OpenAI GPT APIs, and validated Retrieval-Augmented Generation (RAG) pipelines for accuracy and reproducibility. • Deployed ML services via Docker and Kubernetes, establishing CI/CD QA gates for unit testing, integration testing, and automated deployment validation to ensure scalab
- Data ScientistTCSJan 2019 - Oct 2023 (4 years 10 months)• Designed and deployed an end-to-end Machine Learning platform that processed 25M+ financial transactions per hour, enabling real-time fraud detection and risk analysis. • Built ML classifiers and deep learning models (CNNs, RNNs, LSTMs) in Python, TensorFlow, and Scikit-learn, achieving 87% fraud detection accuracy and reducing false positives by 30%. • Leveraged Generative AI and Large Language Models (LLMs) with Transformers and spaCy to summarize unstructured financial data, generate compliance alerts, and enhance anomaly explanations for stakeholders. • Engineered cloud-native ETL/ELT pipelines with AWS Glue, Azure Data Factory, and Snowflake, migrating legacy systems and improving financial data availability by 45% for near real-
Education
verified_user
0% verified
- BBuilding LLM Applications with Prompt Engineering (Link)Apr 2025 - Current (1 year 5 months)
- Masters of ScienceUniversity of New HavenAug 2023 - May 2025 (1 year 10 months)
- MMicrosoft Certified: Azure Fundamentals (Link)Jan 2023 - Current (3 years 8 months)
- GGoogle Cloud Certified: Associate Cloud Engineer (Link)Jan 2022 - Current (4 years 8 months)
- Bachelors of TechnologyJawaharlal Nehru Technological UniversityMay 2016 - Jul 2020 (4 years 3 months)
Projects (professional or personal)
verified_user
0% verified
- RRemote Patient Care: Health Monitoring S Adherence (AWS S IoT)Jan 2025 - May 2025 (5 months)• Developed an IoT-based remote patient monitoring system using wearable devices integrated with AWS IoT Core and IoT Analytics, enabling real-time tracking of patient vitals such as heart rate, blood pressure, and oxygen levels. • Designed a data pipeline with AWS Glue, SageMaker, and CloudWatch to ingest and analyze continuous health readings, generating early risk alerts that improved clinical decision-making and reduced emergency admissions. • Implemented predictive analytics for personalized care pathways, identifying early signs of non-adherence and sending automated medication reminders, which improved adherence rates by 25%. • Integrated the solution with HIPAA-compliant cloud storage and anonymization protocols, ensuring patien
- CChatbotJan 2025 - May 2025 (5 months)• Developed an AI-powered chatbot using Retrieval-Augmented Generation (RAG) and fine-tuning techniques to address university-related queries. Researched RAG methodologies and applied findings to enhance the chatbot's retrieval accuracy and user interaction experience.
- RRadiology Report Generation (Deep Learning)Jan 2025 - May 2025 (5 months)• Built a multimodal deep learning pipeline using BioGPT and CheXNet to generate radiology reports from chest X- ray images. • Processed 7,000+ image-report pairs from the IU Chest X-ray dataset, segmenting reports into Indication, Findings, and Impression sections for training. • Extracted image embeddings via a fine-tuned CheXNet (DenseNet-121) model to enhance clinical feature representation. • Fine-tuned transformer models (T5/BioGPT) using PyTorch and Hugging Face Transformers for medical text generation. • Evaluated model output using BLEU, ROUGE, and BERTScore to ensure clinical accuracy and semantic alignment.