software engineer with AI/ML integration experience.

github

Janardhan Reddy Guntaka  • Peerlist

-AI systems in healthcare can answer anything, but they don’t understand who should see what. This makes them risky in environments where data privacy is critical. -Aegis Health solves this by controlling what AI is allowed to reveal. It masks sensitive patient data before it ever reaches the model. -Same intelligence, but with enforced privacy boundaries. This is how AI becomes usable in regulated systems like healthcare.

Aegis_Hospital_Privacy

Fitness Tracker Microservices Platform Built a production-ready, secure fitness tracking ecosystem from the ground up using modern microservice architecture. Designed and implemented user registration, activity logging, and real-time AI workout recommendations, integrating the Gemini API for intelligent, personalized guidance. Authentication & Security: Implemented robust OAuth2 authentication flow with Keycloak + PKCE, ensuring secure, scalable user access. End-to-End Data Flow: Connected PostgreSQL (user data) and MongoDB (activity logs) across Spring Boot microservices, with real-time inter-service communication via RabbitMQ. AI-Driven Recommendations: Developed an AI-service leveraging Google Gemini API, enabling instant, user-specific fitness advice based on activity patterns. Distributed Queuing & Service Discovery: Integrated RabbitMQ for asynchronous, resilient activity queuing, and Eureka for dynamic registration and discovery of all microservices, boosting scalability and fault tolerance. Modern Frontend: Crafted a fast, responsive frontend in React + Vite with PKCE-based login and intuitive activity dashboards. API Gateway: Centralized security, token propagation, and smart routing across all APIs for a unified developer and user experience. Built to demonstrate expertise in scalable backend systems, secure authentication, real-time AI integration, robust inter-service communication, and full DevOps pipelines. Easily extensible for monitoring (Prometheus) and future cloud deployments. Core Tech: Spring Boot (Java),  React, RabbitMQ, Keycloak, PostgreSQL, MongoDB, Google Gemini API, Docker, Eureka

AI Fitness-Tracker Microservice

🎯 Project Goal Micro-mobility systems like Citi Bike are highly sensitive to rapid shifts in demand. **Accurate short-term forecasts** help operators: - Rebalance bike docks efficiently - Plan proactive maintenance - Maximize bike availability for riders This project ingests historical trip data, aggregates it into hourly time series, engineers sliding-window features, and trains a LightGBM regressor to **predict number of rides in the next hour**. All features and trained models are tracked in [Hopsworks](https://www.hopsworks.ai/) for reproducibility and to enable live inference. An interactive **Streamlit dashboard** visualizes forecasts and station trends.

hourly-citibike-demand-predictor

🎯 Project Goal Traditional taxi services in urban areas like NYC are sensitive to fast-changing demand. **Accurate short-term forecasts** help operators: - Optimize fleet allocation and dispatching - Reduce passenger wait times - Improve operational efficiency This project ingests historical taxi trip data, aggregates it into hourly time series, engineers sliding-window features, and trains a LightGBM regressor to **predict number of rides in the next hour**. All features and trained models are tracked in [Hopsworks](https://www.hopsworks.ai/) for reproducibility and live inference. An interactive **Streamlit dashboard** visualizes forecasts and location trends.

hourly_nyc_taxi_estimation

I had been Team Leader for a hackathon in which our goal is to create a full stack Placement Management System. In a span of 6 months, along with my teammates we have learnt React, Mysql, Django, and Rest services and successfully presented our work in Final Review. This gave me a great experience on Time Management and effectively collaborate with Team, and organize our Roles on Project contribution.

Placement Management System.

Predicting Real estate prices using Linear re

🚀 Real-Time Chat App | MERN Stack + Socket.io A secure, scalable full-stack chat application with real-time messaging, group channels, and JWT-based authentication. Built with React (Vite), Node.js, Express, MongoDB, Zustand, and TailwindCSS.

RealTime Chat Application

Designed and implemented a microservices-based startup idea validation platform using Java 21, Spring Boot, PostgreSQL, Kafka, and Docker, supporting structured idea posts, category management, event-driven architecture, and containerized local development.

startup-validation-platform

Holiday Tick List A simple task management application to help you keep track of your holiday to-dos. This project is built using the MERN stack (MongoDB, Express.js, React, Node.js).