Saurabh Salunkhe

Saurabh Salunkhe

Helping organizations unlock growth through AI, ML, and cloud-powered strategies

Contact Me

About Me

I’m passionate about the transformative power of AI and Generative AI to create real-world impact. At the forefront of this revolution, I specialize in analyzing IT infrastructures, solving business challenges with cloud and AI technologies, and leading POCs, workshops, and technical demos. With 8+ cloud certifications (AWS, GCP, Azure, OCI) and expertise in databases, machine learning, and Generative AI, I design and deploy scalable, AI-enhanced solutions. I’m also exploring Oracle Database Vector capabilities to drive performance and data strategy. From cloud migrations to smart contracts, my work blends technical depth with a future-ready mindset. Backed by a Master's from Northeastern University and a Computer Engineering degree from Mumbai University, I thrive on innovation, adaptability, and building what's next. Let’s build the future—one breakthrough at a time.

My Projects

RAG chatbot with OCI Generative AI service and Oracle Database Vector Store

project name
New

Accessing insights from vast data is challenging—RAG simplifies it. This workshop builds a RAG-based chatbot using Oracle Database 23ai and OCI Generative AI, enabling users to chat with unstructured data like PDFs, CSVs, and TXTs for dynamic, intelligent interaction.

View on GitHub
project name

Fine-Tuning DeepSeek-R1-Distill-Qwen-1.5B for Healthcare Discharge Summaries

This project outlines high-level steps to fine-tune the DeepSeek-R1-Distill-Qwen-1.5B large language model specifically to generate patient discharge summaries tailored for healthcare settings.

View the website

See on Github

project name

Demontization Software

In case a country decides to demonetize its currency, that country should make sure this transition flow smoothly. The exchange of currency process should be optimized and should take into cosideration the different roles involved in this process such as comman man, Local Bank Managers, Federal Bank and Logistics. The application that we are making aims at making this process as smooth as possible.

View on GitHub

project name

Predicting Click Through Rate for online ads

In this project, I have stated the importance of the parameter click through rate in the ads framework model and how significant it is in the revenue generation and relevancy for both the publisher and advertiser. Further, I have demonstrated the current models being used by tech giants like facebook and Yahoo. Then, I have applied different models on the Avito dataset for comparative analysis of all the state of the art machine learning and deep learning model and state its advantage and disadvantages.

View on GitHub

project name

JobsNinja - A Job Portal

This project helps applicants find the apt job position they are seeking for. The applicant can search for the proper jobs and accordingly apply for the jobs and get response from the hiring manager. The applicant can further upload his resume on the jobs site which can be rendered for view by the employer. Users, employers will have their accounts and can perform their tasks accordingly.

View on GitHub

Other Projects

Framingham Heart Study Application

Framingham is the largest town in Mass with a diverse computing of inhabitants. The Framingham Heart Study is a world famous institute that tracks heart disease in a population of 5000 people with their children and grand-children. All of the world-wide advances in the prevention of heart-disease are the result of this study.We develop qualify-of-life metrics that will speak to cardiac health of individuals.Assumption is that individuals, children, and grand-children will use their mobile devices to record their lifestyle and environmental conditions in their surroundings. The first measurements to include will be the vital signs of the individuals in the study and assume that will be automatically recorded though the individual’s mobile device. Calculation of risk score has also been done

View on GitHub

Intrusion Detection System

This is a very affordable home security alarm which overcomes the limitations of the already existing burlgar alarms and at the same time can be manufactured very cheaply. All bodies radiate heat in the form of infrared which is not visible to human eyes. However, this can be detected by electronic sensors like PIR sensor. This sensor generates electric charge when exposed to heat and sends a signal to arduino. According to level of the infrared in front of sensor, the GSM model sends a SMS and call to the ownner alerting him about the situation.

View on GitHub

Discuss - An idea management application

Everyone holds the potential to learn coding and master any programming language. There are many people who wants to delve into coding and this website provides a platform for all new aspiring programmers to discuss about coding and how to bridge the gap between ideas and technical aspects by involving into disussions with expert coders in the industry.

View on GitHub

Work Experience

Software Engineer Intern - Logitech (Jun 2017 - Aug 2017)

• Worked closely with the IT Director to apply Machine Learning to JIRA data-sets to enhance user experience
• Used bag of words approach with TF-IDF and applied k-nearest neighbor algorithm with cosine similarities
• Successfully predicted JIRA ticket label and assignee and helped reduce manual efforts on global help desk
• Developed chat based user interface using jQuery, CSS3 and used jira-python client for automated tickets
• Integrated a flask server and created REST APIs endpoint and technical documentations for the same
• Proposed and prototyped idea of creating JIRA tickets with speech recognition with suggested label/assignee

Research Analyst -Directi (2015 - 2016)

• Presented identified patterns about domain names through a web-based dashboard (D3.js) to senior executives
• Revamped UI/UX of domain parking pages by studying click patterns and made relevant web template for ads
• Achieved more than 200% rise in revenue of multiple clients’ web portfolio by deploying targeted ads
• Side project included creating a machine learning model to predict click estimation of context ads

Extra_Curriculars