Manish Dhruva Email & Phone Number

Dubai, Ae

Washington, United States

◦ 𝗣𝗶𝗽𝗲𝗹𝗶𝗻𝗲 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻: Replaced Java-based patent concept extraction with Python and ML models (FAISS, Mistral-7b), reducing complexity by 90% and enhancing quality.◦ 𝗜𝗻𝗱𝗲𝘅 𝗖𝗿𝗲𝗮𝘁𝗶𝗼𝗻: Created FAISS index with 90M+ embeddings, empowering semantic search applications and potentially expanding customer base by 50%.◦ 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 𝗘𝗻𝗵𝗮𝗻𝗰𝗲𝗺𝗲𝗻𝘁: Developed FAISS index update scripts for weekly automation, ensuring system efficiency and relevance.◦ 𝗔𝗣𝗜 𝗜𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲: Designed Django APIs for frontend-backend communication in Relecura's applications, promoting UI-Script integration.

Bengaluru, Karnataka, India

◦ 𝗪𝗲𝗯 𝗦𝗰𝗿𝗮𝗽𝗶𝗻𝗴: Developed an automated Javascript scraping code to extract data from 36,500 webpages from the VCCEdge repository, replacing manual data collection and saving 100 days of laborious work.◦ 𝗗𝗮𝘁𝗮 𝗪𝗿𝗮𝗻𝗴𝗹𝗶𝗻𝗴: Performed end-to-end data processing including discovery, structuring, cleaning, feature engineering, and validation of fund & company records sourced from scraped webpages.◦ 𝗥𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗦𝘆𝘀𝘁𝗲𝗺: Developed content-based recommendation system using streamlit to match potential investors with suitable funds and identify potential funds for sale, facilitating informed investment decisions.

Bengaluru, Karnataka, India

◦ 𝗧𝗶𝗺𝗲 𝗦𝗲𝗿𝗶𝗲𝘀 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴: Implemented advanced forecasting techniques like ES, SARIMAX, XGB-Regressor to accurately project solar power generation, achieving a 93% accuracy rate in mitigating solar intermittency.◦ 𝗢𝗯𝗷𝗲𝗰𝘁 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗦𝗲𝗴𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻: Executed PV panel instance segmentation and identified faulty panels using detectron2/yolov5 and yolov3-tiny models, improving solar array analysis.◦ 𝗗𝗮𝘁𝗮 𝗔𝗻𝗻𝗼𝘁𝗮𝘁𝗶𝗼𝗻: Accomplished image annotation of 5000 solar arrays and 1500 airport runway images within a tight timeframe of 4 days, utilizing efficient tools like LabelMe and CVAT for accurate and detailed annotations.

Chennai, Tamil Nadu, India

◦ 𝗠𝗼𝗱𝗲𝗹 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴, 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴, 𝗧𝗲𝘀𝘁𝗶𝗻𝗴 & 𝗙𝗶𝗻𝗲-𝗧𝘂𝗻𝗶𝗻𝗴: Built a custom CNN model for accurate recognition of English handwritten manuscripts with a 95% accuracy rate. Additionally, trained a domain-specific spaCy-NER model for effective identification and visualization of named entities in clinical text.◦ 𝗗𝗮𝘁𝗮 𝗔𝗻𝗻𝗼𝘁𝗮𝘁𝗶𝗼𝗻: Performed precise and comprehensive text annotation of entities such as chemicals, dosage, diseases, and symptoms on PubMed abstracts using the powerful Prodigy tool, facilitating accurate information extraction and analysis from scientific literature.

Master'S Degree, Data Science & Machine Learning, 9.07

Bachelor Of Engineering - Be, Computer Science, 6.5

VTU Syllabus

Secondary School, Pcmc, 70.22%

CBSE Syllabus

High School, Pcmc, 87.33%

ICSE Syllabus