Kunj Patel

Dynamic software developer with a proven track record at Oasis Infobyte, adept at crafting responsive web applications and integrating APIs. Recognized for problem-solving skills and collaboration in cross-functional teams, leveraging expertise in JavaScript and REST API design to deliver impactful solutions that enhance user experience.

• Multi-Agent Reinforcement Learning for V2X Communication, Independent Project, Designed a multi-agent reinforcement learning simulation in Python where autonomous agents plan and act under shared state incorporating Mixed-Integer Linear Programming (MILP) agent analysis and a custom FITS neural-network backbone.
• Bayesian Optimization for Multi-Objective DNN Hyperparameter Tuning, Group Project, Collaborating with a 4-person cross-disciplinary team to build a multi-objective optimization framework (Bayesian Optimization + NSGA-II), translating an evolving problem scope into a structured technical report approved by faculty for IEEE-style publication.
• Real-Time Smart Grid Anomaly Detection Pipeline, Group Project, Built a real-time data pipeline (Apache Kafka + LSTM + Django) with a live monitoring dashboard, owning feature engineering, data-quality analysis, and error-handling across the stack.
• Full-Stack Web Applications & API Integrations, Openstream NFT Marketplace, Built full-stack applications integrating third-party APIs (GitHub API, Thirdweb) for real-time data sync and on-chain storage direct experience designing and consuming external service integrations.
• Real-Time Fraud Detection Pipeline, Independent Project, Designed and tested a real-time fraud-detection pipeline using Apache Spark Structured Streaming with session-window aggregation, verifying correctness end-to-end against flagged fraudulent sessions.
• IoT Platform Engineering: MQTT/CoAP Pipeline, Coursework Project, Built an IoT device-communication pipeline using MQTT and CoAP protocols, achieving 18/18 passing tests hands-on experience integrating and orchestrating multiple connected systems and data flows.
• Time Series Forecasting & Anomaly Detection, Coursework Project, Built and compared ARIMA, Random Forest, and Gradient Boosting models for traffic time-series forecasting and anomaly detection on the PeMSD7 dataset.