M
Megha Ghosh
About
Detail
Karnataka, India
Contact Megha regarding:
work
Full-time jobs
Résumé
Jobs
verified_user
0% verified
GenAI EngineerClouderaJun 2025 - Current (1 year 2 months)• Architected and deployed an enterprise-grade Generative Al- powered HR Assistant using Retrieval-Augmented Generation (RAG) to answer policy-related employee queries with high accu- racy and interpretability. • Engineered a semantic search pipeline using OpenAl text embed- dings, FAISS vector database, and custom ranking algorithms, en- abling contextual document retrieval from large corpora of HR pol- icy documents. • Implemented advanced prompt engineering techniques and in- tegrated multi-shot examples with metadata-enriched context to minimize hallucination and improve LLM reliability. • Developed modular microservices for document preprocessing, vectorization, query embedding, and real-time retrieval, orches- trated using AWS Lam- Data Engineer II - AI SystemsMicronJul 2023 - Mar 2025 (1 year 9 months)Designed and implemented an AI-powered knowledge intelligence platform integrating enterprise data pipelines with LLM-based retrieval systems for intelligent document search and workflow automation. • Built scalable ETL pipelines using Apache NiFi, Python, and SQL to ingest and process structured (MSSQL, MySQL, SAP HANA) and unstructured data, handling high-volume datasets across multiple sources. • Implemented Retrieval-Augmented Generation (RAG) pipelines using FAISS vector database and embedding-based semantic search for context-aware response generation. • Developed backend services and REST APIs integrating data pipelines with AI models, while designing hybrid re- trieval strategies that improved accuracy by 25%. • Optimized data proce
Education
verified_user
0% verified
- NMaster of TechnologyNational Institute of Technology (NIT)Jan 2021 - Jan 2023 (2 years 1 month)
- IBachelor of Technology [B.Tech]Indian Institute of Information Technology (IIIT)Jan 2016 - Jan 2020 (4 years 1 month)
Projects (professional or personal)
verified_user
0% verified
- OOrgan at Risk Image Segmentation Using TransformerJan 2022 - Jan 2023 (1 year 1 month)• Developed a hybrid deep learning model combining Vision Trans- formers and U-Net for accurate segmentation of Organs at Risk (OAR) in medical images. • Focused on enhancing segmentation precision in CT scans to sup- port radiation therapy planning and reduce manual intervention by radiologists. • Improved model performance using data augmentation, patch- based learning, and attention mechanisms for better spatial under- standing. . Evaluated results using metrics like Dice Coefficient and loU, demonstrating improved accuracy over traditional CNN-based methods. • Aimed at enabling automated and reliable clinical decision sup- port systems in oncology workflows.
- TTwitter Sentiment AnalysisJan 2019 - Jan 2020 (1 year 1 month). Built a sentiment analysis pipeline to classify tweets related to the Indian demonetization event as positive, negative, or neutral. • Utilized Apache Hive for large-scale tweet data storage, prepro- cessing, and querying to derive sentiment trends from unstructured data. • Applied basic NLP techniques like text normalization, tokeniza- tion, and polarity scoring to analyze sentiment. • Designed Hive queries to compute sentiment distributions over time, offering insights into public opinion during socio-economic events. • Gained hands-on experience in big data analytics, HiveQL, and real-world text data processing.