Work Experience

Software Engineer, Associate Technical Consultant - Perficient

• Implemented a custom software middleware layer to modernize, automate and integrate warehouse management system by utilizing Twelve-Factor App methodology.
• Integrated WMS with ERP, SQL Server and other internal services which simplifed the returns and slahsed the warehouse management cost significantly.
• Designed, developed and implemented RESTful webservices for the distributed application using Java Spring boot and deployed as Docker images.
• Collaborated with stakeholders to identify product and technical requirements to facilitate appropriate solution design.
• Created PCOs to validate architectural designs and improvements.
• Worked with the QA team by debugging and troubleshooting code to ensure applications met specifications and requirements.
• Collaborated closely with peers to build seamless APIs, libraries and producing client deliverables such as detailed design documentation, test plans, and well-documented code.
• Collaborated with project stakeholders to identify product and technical requirements to facilitate the most appropriate solution design.
• Conducted code analysis to determine integration needs and enhance complex software components.
• Practicing strong configuration management and version control to manage software development.
• Experience working in an Agile/Scrum environment and collaborate with cross-functional teams to tackle customer’s problems.

IT & Web Services - Auraria Library, University of Colorado Denver

• Developed, maintained, and improved Auraria Library website which served for tri-institution campus.
• Implemented Global Menu Redesign and Map Redesign with a focus on HTML, JavaScript, jQuery, and React.
• Collaborated with cross-functional departments to design and develop intuitive user interfaces and components.
• Planned, designed, and conducted research work to assess web services to identify user needs.
• Achieved 50% reduction in deployment time by redesigning batch deployment process.

Academic Projects

Human-Level Performance Control Through Deep Reinforcement Learning

Technologies : Python, NumPy, Pandas, TensorFlow, Keras, CNN, Deep Q Learning, Open GYM AI.

• Proposed Convolutional Neural Network Model with Deep Q-Network to play an Atari game by learning control policies and solving Markov Decision Process by observing from raw input(emulator).
• Calculated accuracy based on outcome rewards and presented the model achieves better performance than a human player.

Generating Text Using Recurrent Neural Network

Technologies: Python, NumPy, Pandas, TensorFlow, Keras, RNN, Matplotlib

• Developed and Implemented Long Short-Term Memory Neural Network to generate the text by sampling the next character based on probability distribution using Python and Keras.
• The model has achieved 70% accuracy and generated a new sequence of characters based on random seeds of text.

Crime Prediction using Machine Learning Models

Technologies : Python, NumPy, Pandas, SQL, Spark, MongoDB, Google Cloud, Scikit-learn, Matplotlib

• Developed and compared accuracy of different machine Learning algorithms like Naïve-Bayes and Decision Tree to perform classifications on a huge crime dataset to predict future crime types and locations based on the crime locations, time and types.
• Performed visualization on the dataset and enhanced the accuracy to 82%.

Integration of Big Data and Cloud Computing to detect money laundering using RAQ

Technologies: Java Sping, Hibernate, Hadoop, Spark, SQL, Google APIs, Fast RAQ

• Designed and compared different time-series machine learning models by adopting Random Forest and Bayesian Network to predict the “hot spot” crime locations.
• • Analyzed patterns to predict the crime types given the crime locations, time, and types. Enhanced the accuracy to 76% and performed visualization using google maps.

Let's Get In Touch!

Give me a call or send me an email and I will get back to you as soon as possible!

Phone: (610)620-3015

Email: Madhan Vibeeshanan

Github LinkedIn