Machine Learning Software Engineer

Spoorthi Subramanya Bhat

+1-774-***-**** ********************@*****.*** Linkedin Github

Clayton, MO(Open to Relocate)

OBJECTIVE

Results-driven Backend Engineer with 3+ years of experience in Android development, Software Engineer, and AI/ML solutions across healthcare, cybersecurity, and e-commerce. Expertise in building scalable applications, data-driven systems, and production-ready ML models with strong skills in Kotlin, Java, Python, and cloud technologies. Proven record of delivering impactful solutions supported by publications, awards, and cross-functional collaboration. Actively seeking full-time opportunities to contribute technical expertise, innovation, and measurable business value. PUBLICATIONS AND AWARDS

• Automated Medical Coding Using a Hybrid Decision Tree with Deep Learning Nodes: Spoorthi Bhat, Haiping Xu, Joshua Carberry.Proceedings of the 11th IEEE BigDataService 2025, Tucson, AZ, USA, July 21–24, pp. 81–88. H.A.

• Indoor navigation using BLE beacons: In Recent Trends in Computational Sciences, 1st Edition, CRC Press, 2023. ISBN: 978**********, By Chaya Kumari, Spoorthi S. Bhat, M. Sucharith, T.S. Pooja, B.C. Thanmayi Awards

• FUCHS Intelligent Data Discovery Hackathon 2024 - Won the ’Most Impactful Solution’ and ’Best Machine Learning Technique’ awards ($500 total)

PROFESSIONAL EXPERIENCE

University of Massachusetts ( Various Roles ) Jan 2024 - May 2025 Role 1: Backend Engineer Dartmouth, US

Client: College of Nursing and Health Sciences, UMass Developed a comprehensive app designed to support individuals on their journey to recovery from Substance Use Disorder, leading the translation of nursing team requirements into an intuitive, feature-rich mobile solution.

• Led end-to-end development from requirements gathering to deployment, delivering user-focused features that boosted engagement by 80% and reached 150+ active users

• Developed and maintained Java applications using Spring Boot and Microservices architecture, delivering scalable and modular backend solutions.

• Designed and consumed REST APIs and integrated with Oracle and PostgreSQL databases, ensuring high-performance data access.

• Built and deployed applications in cloud environments, leveraging Jenkins CI/CD pipelines, Kafka messaging, and automation tools for efficient delivery.

• Contributed to open-source software and implemented best practices for code quality, collaboration, and maintainability.

• Designed and implemented core modules, including daily check-ins, sobriety counter, goal setting, journaling, and peer support using Jetpack Compose, MVVM, and Kotlin, ensuring a responsive and accessible user experience

• Enhanced application usability and navigation by iterative UI/UX improvements and rapid incremental releases, cutting feature delivery timelines by 30% and improving retention rates within the first 3 months. Role 2: Machine Learning Engineer

Client: Computer Science Department

Developed a machine learning pipeline to automatically classify National Vulnerability Database (NVD) records with the correct Common Weakness Enumeration (CWE) code, leveraging large language models to improve accuracy in software vulnerability tagging.

• Led experimentation with multiple LLM architectures for automated CWE classification, evaluating performance on 12k+ NVD records and selecting BERT with a fine-tuned classification head as the optimal solution

• Experienced in model deployment and optimization for production environments.

• Skilled in Natural Language Processing (NLP) techniques and applications.

• Proficient with PyTorch for model training and experimentation.

• Strong foundation in machine learning frameworks for building scalable solutions.

• Designed and implemented the data processing pipeline, ingesting NVD JSON feeds, extracting CVE descriptions via the CVE API, and preparing inputs for model training using PyTorch

• Achieved a 22% improvement in classification accuracy through hyperparameter tuning, model optimization, and systematic evaluation against baseline approaches. Role 3: Machine Learning Engineer

Client: Computer Science Department

Developed an underwater acoustic target recognition system to help marine researchers automatically identify vessel types from sonar recordings, reducing the need for manual audio inspection and enabling faster, more accurate analysis.

• Addressed the challenge of manual vessel classification by processing and transforming audio signals from ShipsEar and DeepShip datasets using FFT, STFT, and DCT for precise frequency-domain representation

• Applied advanced feature extraction methods (PCA, Mel-Spectrograms, ICA) to enhance model input quality and improve classification robustness in noisy underwater conditions

• Built and trained a CNN ResNet-50 model for multi-class vessel classification, achieving 87% accuracy and deploying it as part of the research workflow.

IN TIME TEC March 2022 - Aug 2023

Software Engineer Bangalore, India

Client 1: Quintet - CALS

Contributed to the development of a scalable no-code/low-code website-building platform, engineering backend integrations, and database workflows to seamlessly connect drag-and-drop widgets with complex relational data in real time.

• Built robust backend data integrations, writing complex SQL queries, triggers, and stored procedures to power dynamic page rendering within the visual builder

• Designed and optimized workflows across multiple relational databases, improving query performance by 20% and ensuring seamless data retrieval for end-user-generated sites

• Automated repetitive backend processes, reducing manual interventions and enabling real-time widget-to-database connectivity.

Client 2: Kount, an Equifax Company

Developed a secure, modular Android SDK to capture device and behavioral data for fraud detection, enabling bot vs. human classification in real time across 100K+ production devices.

• Developed an Android SDK supporting Kotlin, Java, and cross-platform frameworks to collect device sensor activity, location, and behavioral patterns for fraud analytics

• Boosted bot detection accuracy by 25% by enabling real-time data flow to AI/ML fraud detection models, improving proactive threat identification

• Reduced integration issues by 40% through modular SDK architecture, well-structured APIs, and comprehensive client-facing documentation.

Client 3: DripDropDistro

Developed the Android version of an existing iOS e-commerce application from scratch, responsible for backend feature implementation and deployment to the Google Play Store.

• Built and integrated scalable backend features to support product browsing, ordering, and payment flows, ensuring smooth interoperability with existing iOS and web platforms

• Reduced API call redundancy by 30% through smart caching strategies, improving performance and reducing server load

• Enhanced session security with encrypted local storage and secure token handling, strengthening protection against unauthorized access.

SVACHALLAN SOLUTIONS Jan 2022 - March 2022

Android Development Intern Bangalore, India

Contributed to the design and development of Helping Hand, an Android application connecting users with community assistance services such as volunteering, donations, and local help requests.

• Designed and implemented a visually appealing Java-based UI, ensuring intuitive navigation and responsive layouts

• Integrated Firebase Authentication for secure and reliable user login and account management. EPSILON SCIENTIFIC June 2020 - Aug 2020

Website Development Intern Bangalore, India

Contributed to the redesign and enhancement of an existing social networking web application to improve usability and performance.

• Restructured and optimized the application’s architecture, improving user experience and maintainability

• Developed and updated features using HTML, CSS, PHP, JavaScript, and MySQL, delivering dynamic and responsive web functionality.

PROJECTS

ReLeaf – Android Application for Substance Use Disorder

• Developed an end-to-end Android application using Jetpack Compose, Room, and MVVM architecture to enable goal tracking, daily check-ins, and journaling for SUD patients, featuring offline access with seamless real-time synchronization via Firebase

• Integrated push notifications, encrypted local storage, and secure authentication flows, enhancing user engagement, safeguarding sensitive health data, and ensuring compliance with healthcare security standards (e.g., HIPAA). Automated Medical Coding Using Hybrid Decision Trees - Master’s Thesis

• Designed and implemented a hybrid decision tree classification pipeline combining rule-based decision trees and LSTM models to automate ICD code prediction, reducing inference time by 38% on clinical discharge summaries

• Achieved a 17% improvement in multi-label classification accuracy over deep learning-only and full rule-based baselines by applying hierarchical modeling and leveraging GPT-4 for semantic sentence extraction. Weather Forecast – Modern Android App (Personal Project)

• Built a weather forecasting app using MVVM architecture, Repository pattern, and Dagger-Hilt to keep the code modular, easy to test, and scalable

• Connected real-time weather APIs with location services, added rate limiting, and created fallback options to keep the app stable even with slow networks or API limits

• Improved UI responsiveness by 35% by optimizing Jetpack Compose rendering, using lifecycle-aware coroutine flows, and reducing cold-start time with asynchronous state handling. Personal Expense Manager – Full Stack Web Application (Personal Project)

• Designed and developed a full-stack expense manager with Spring Boot (REST APIs), JavaScript frontend, and MongoDB backend, supporting multi-account tracking and CRUD operations

• Applied NoSQL schema design in MongoDB to manage user-account-expense relationships efficiently.

• Implemented secure session-based authentication, modular dashboard components, and client-side search/filtering to enhance usability and system scalability.

VQA for Radiology – Medical QA System Using Vision-Language Models (Personal Project)

• Developed a radiology-focused visual question-answering (VQA) system by fine-tuning LLaVA-Med on the VQA-RAD dataset using LoRA. Combined computer vision and NLP to assist radiologists with scan interpretation, improving diagnostic support, and reducing time-to-decision. Sports Management Client-Server Java Application (Personal Project)

• Built a Java-based client-server system using socket programming for real-time sports team and game management

• Developed a multi-threaded server with JSON-based communication and MySQL integration for persistent data handling.

Technical Skills

• Databases: SQLite, MySQL, Firebase, MongoDB, Room Database

• Programming Languages: Kotlin, Java, Python (NumPy, Pandas, Matplotlib, Scikit-learn, PyTorch), JavaScript (d3.js)

• Android Development: Jetpack Compose, Compose Multiplatform, MVVM, Retrofit, Hilt, WorkManager, Jetpack Navigation, Firebase Realtime Database, Room, ML Kit, Android SDK

• Tools and Libraries: Git, GitHub, GitLab, Bitbucket, Postman, Android Studio, IntelliJ IDEA, VS Code, Gradle, Jira EDUCATION

UNIVERSITY OF MASSACHUSETTS Dartmouth, US

MS in Computer and Information Science (GPA: 4/4) Sep 2023 - Aug 2025 VISVESVARAYA TECHNOLOGICAL UNIVERSITY Mysuru, India B.E. in Computer Science (GPA: 3.7/4) Aug 2018- Aug 2022

Contact this candidate