Bhagyesh Rathi
SOFTWAREDEVELOPER& AI/MLENGINEER
About Me
I am a Software Developer and AI/ML Engineer with a passion for building intelligent, scalable systems. Currently pursuing my Master's in Artificial Intelligence at San Jose State University, my focus lies at the intersection of robust backend engineering and cutting-edge machine learning.
At Rakuten, I engineered high-impact microservices, implemented secure OAuth 2.0 architectures, and orchestrated GCP deployments with Kubernetes. Whether it's developing interactive RAG pipelines, optimizing distributed systems, or training predictive models, I thrive on turning complex technical challenges into seamless user experiences.
When I'm not writing code, you can find me exploring the latest advancements in LLMs or refining my problem-solving skills.
Experience
Software Engineer Intern
Rakuten · San Mateo, CA
- Engineered Kotlin microservices for Social Authentication (Google, Apple, Facebook), driving a 40% adoption rate and a 37% increase in conversion rate.
- Developed robust REST APIs utilizing OpenID Connect and OAuth 2.0 to implement secure token-based authentication and authorization flows
- Architected scalable infrastructure on Google Cloud Platform (GCP), orchestrating Docker containers with Kubernetes via automated GitLab CI/CD pipelines
- Established comprehensive system observability by integrating OpenTelemetry with GCP Cloud Trace and structured Log4j logging for real-time performance insights
Software Engineer Co-op
Rakuten · San Mateo, CA
- Architected and deployed a core internal SDK that abstracted complex service-to-service communications, successfully published to Artifactory via GitLab CI/CD
- Engineered secure event listeners using Kotlin to intercept and process critical compliance signals (consent revocation, account deletion) from OAuth providers
- Implemented robust security protocols by validating and decoding JSON Web Tokens (JWT) to securely authenticate inter-service requests
RESEARCH INTERESTS
Geometric Consistency: Latent Space Pruning for Chain-of-Thought Reasoning (In Progress)
This paper proposes and evaluates a novel unsupervised method, Geometric Consistency, designed to enhance reasoning reliability by filtering outliers within the latent vector space.
AutoML (In Progress)
Automated Machine Learning
Projects
RAG Based Interactive Resume
Implemented a production-grade RAG deployment on a user-facing portfolio site
- Ingestion: Parsed and chunked resume PDF into semantic segments
- Embedding: Embedded text chunks using a transformer model to generate dense vector representations
- Storage & Retrieval: Stored vectors in a Pinecone vector database and implemented retrieval to compare user queries against stored vectors
- Generation: Passed retrieved context + query to Vertex AI (Gemini) LLM to generate accurate, context-aware responses
AutoML
FullStack AutoML platform and code generator
- Engineered an AutoML system using Scikit-Learn & Flask that automates preprocessing, task detection, and parallel model training/evaluation
- Designed a responsive React frontend with real-time interactive ROC/Scatter plots and confusion matrix for visualizations of top model
- Reduced training latency by 40% implementing Stratified Sampling and dynamic model switching to handle large datasets efficiently
- Developed a context manager to profile real-time CPU/RAM usage & a leaderboard to sort models by accuracy, time, or resource efficiency
- Built a transpiler engine to enable one-click downloads for both serialized models (.pkl) and their reproduction code
Bank Churn Data Analysis and Prediction using ML
Data analysis and prediction using ML
- Analyzed data of 10,000 account holders at a Multinational Bank by doing exploratory data analysis with Pandas
- Constructed a streamlined pipeline for training 5 machine learning models to predict customer churn using Scikit-Learn
- Implemented XGBoost, Random Forest, KNN, SVM, and Naive Bayes models and compared them
- Utilized N-fold cross-validation, F1-score, confusion matrix to evaluate the performance of each model
Skin Cancer Detection using CNN
CNN-based image classification for skin cancer detection
- Developed a CNN using Scikit-Learn to classify skin lesion images into 7 cancer categories, achieving 80% accuracy
- Preprocessed data with resizing, normalization, one-hot encoding, and oversampling to address class imbalance
- Optimized model performance using the Adam optimizer, learning rate annealing, and hyperparameter tuning
Skills & Technologies
Education
Masters of Science in Artificial Intelligence
Expected: May 2027San Jose State University · San Jose, CA
Relevant Coursework:
Bachelors of Science in Computer Science
August 2020 — May 2024San Jose State University · San Jose, CA
Relevant Coursework:
Get in Touch
I'm always open to discussing new opportunities, collaborations, or interesting projects.