Machine Learning Computer Vision

VEDANT SAWANT

*******@***.*** +1-814-***-**** linkedin.com/in/vedantsawant6900 github.com/VedantSawant6900 EDUCATION

The Pensylvania State University Aug 2024 – May 2026 Master of Science in Computer Science and Engineering GPA: 3.77 University of Mumbai Aug 2018 – May 2022

Bachelor of Engineering, Computer Engineering GPA: 8.99 WORK EXPERIENCE

Software Engineer Meditab Software Inc. Jul 2022 – Mar 2024

• Collaborated with cross-functional teams to develop the backend of a Drug Dispenser using four interdependent concurrent Finite State Machines (FSM), integrated via REST API and CouchDB, enabling real-time state visualization.

• Engineered an end-to-end image preprocessing FastAPI server integrated with Meta’s Detectron 2 for pill detection and segmentation, enabling frontend visualization and storing data for training and annotation.

• Enhanced a computer vision system for blister pack verification, achieving 100% accuracy.

• Developed a high-precision pill dimension measurement device achieving an accuracy of 0.01 mm.

• Oversaw the creation of an NXP-powered CherryPy server, enabling communication via AWS (MQTT, Lambda) to capture 30 fps videos for validating medication adherence for insurance claims.

• Performed data modelling and established an automated build and deployment pipeline using GitLab Runner, shell scripting, AWS Lambda, S3, MQTT and RESTful API, automatically updating devices on commits to the master branch.

• Built Docker files and images to deploy applications on Kubernetes for healthcare sector in diverse architectures.

• Leveraged multi-threading and multi-processing to optimize concurrency in complex solutions.

• Troubleshooted and resolved critical customer escalations and high-performance scaling bugs effectively. Machine Learning Intern Madras Scientific Research Foundation Jul 2021 – Dec 2021

• Explored and implemented Machine Learning Models for movie ticket pricing prediction with a R2 score of 0.86.

• Developed a computer vision system to filter primary colors with 90% accuracy. SKILLS

• Developer Tools: Hugging Face, Git, Docker, Kubernetes, CI/CD, Amazon Web Services (AWS), REST API, Jenkins, GitLab, Linux, Unix, Postman, Anaconda, MySQL Workbench, DBeaver, Data Structures, Algorithms, Jira, Odoo, Grafana, Loki

• Software Lifecycle: Object-Oriented Design, Test-Driven Development, Agile, Scrum, Verification, Validation, System Integration, Unit Testing, Integration Testing

• Programming Languages, Database: Python, Java, SQL, JSON, MySQL, MariaDB, NoSQL, CouchDB

• Libraries: PyTorch, Keras, TensorFlow, Pandas, Hugging Face, NumPy, Scikit-learn, Matplotlib, Seaborn, OpenCV

• Soft Skills: Innovation, problem-solving, collaboration, communication skills, time management, leadership PROJECTS

MeaCap: Memory-Augmented Zero-shot Image Captioning (Hugging Face, Python, Deep Learning, NLP) Dec 2024

• Validated the original MeaCap paper's approach to zero-shot image captioning as part of an open-source initiative.

• Improved performance by 5% using MPNet V2 and optimized compatibility with broader deep learning frameworks. Deep Learning Classifier for LLM Identification (Python, LLM, DL) Oct 2024

• Generated a dataset of 13,000 AI-generated sentences using LLAMA, GPT, PYTHIA, OPT, and BLOOM to identify the Large Language Model responsible for specific text completions.

• Achieved 48% accuracy with LSTM-RNN and 90% with BERT-classifier in detecting the LLM used for sentence. Handwritten Devanagari OCR (Python, Computer Vision, Deep Learning) May 2022

• Applied computer vision techniques, including Hough Line Detection, Morphological Operations, and Otsu thresholding, to segment individual characters from input images.

• Trained a CNN model on a dataset of 92,000 handwritten Devanagari character images, categorized into 46 classes with an 85-15 split for training and testing, achieving 86% accuracy in identifying single characters. Insurance Charges Prediction (Python, Machine Learning) Dec 2020

• Handled missing values in dataset using mathematics & statistics and visualized the data spread with statistical plots.

• Performed feature scaling and compared the performance of Regression models (up to degree 20) using R2 metric. PROFESSIONAL CERTIFICATIONS

• Microsoft Technology Associate: Introduction to Programming using Python by Microsoft (MTA: 98-381)

• Certified Kubernetes Application Developer (CKAD)

• IBM Data Science Professional Certificate

Contact this candidate