Harbhajan Singh

• Working as a senior data analyst. My key responsibilities are developing scripts for ETL processes, storing data on different database platforms, apps creation in PowerApps, Dashboard creation, etc.

• Providing consultancy and technical guidance to develop an Agentic RAG application using Python, FastAPI, and OpenAI’s LLM models, focusing on scalability and robust performance.

• Assisting developers in designing and implementing efficient RAG pipelines by integrating semantic search, vector databases (e.g., Pinecone, Weaviate), and prompt engineering for enhanced retrieval accuracy.

• Collaborating on API architecture, debugging, and optimization to ensure a seamless user experience while aligning ongoing development efforts with business objectives.

• As a freelance data scientist, I've had the privilege to work with a diverse range of clients, applying my skills in data science and artificial intelligence to solve real-world problems.

• From building predictive models to uncovering actionable insights, I've been dedicated to delivering solutions that drive tangible results and exceed client expectations.

• I'm passionate about leveraging my expertise in data science and artificial intelligence to help clients achieve their goals and achieve their goals.

• Worked on a RAG-based chatbot project, AgPal(https://agpal.ca/en/home), automating test scripts using Azure OpenAI services, including Prompt Flow, Semantic Kernel, LangChain, and Power Automate, to enhance system functionality and efficiency.

• Utilized Prompt Flow to manage project workflows and test the accuracy of LLM responses, ensuring robust chatbot performance based on GroundTruth and Groundness score.

• Developed and implemented new features for a deduplication tool, fixing numerous existing bugs to improve tool reliability.

• Conducted data collection for the detection and classification of arthropods and set the data pipelines by collaborating with a research center to build comprehensive datasets.

• Led the Internal Project Management project, developing a dashboard to monitor various team projects, track project statuses and tasks, and assess team capabilities.

• Led end-to-end development of a RAG-based (Retrieval-Augmented Generation) product for extracting text using OCR techniques from diverse PDF formats, including scanned documents, pertaining to Home Owner Association (HOA) data.

• Using BERT implemented rigorous data correction procedures to ensure accuracy, rectifying spelling errors within extracted text and storing data efficiently in a database.

• Developed a user-friendly RAG and Streamlit-based UI, integrating Large Language Models (LLMs) like GPT-3.5-turbo and Langchain, to provide quick and precise responses to user queries based on the connected vector database as a knowledge base.

• Collaborated with stakeholders and reporting & analytics team at a leading global investment banking and financial services firm (Credit-Suisse) to construct and deploy data-driven solutions using Python, SQL, and other tools & software.

• Designed and accomplished metadata pipelines to efficiently store and process large data sets by improving performance, empowering data scientists to formulate predictive models and forecasts for customer queries and solutions.

• Led & managed value-added projects, including the development of a rule-based Chatbot by utilizing NLP techniques and the creation of two innovative products i.e. SQL checklist matching and JIRA monitoring, and achieved high accuracy and efficiency.

• Applied a range of tools and technologies including Tableau, DataLake, and Azure Clouds to perform exploratory data analysis, feature engineering, and model training and evaluation for a variety of business use cases.

• Applied Time Series forecasting to predict the number of requests and reports raised and requested by users & customers.

• Leveraged AI and ML to produce reports, resulting in an 85% reduction in manual efforts for business analysts and customers.

• Enhanced and optimized data processing and data modeling workflows to ensure maximum efficiency (increased efficiency by 90%) and accuracy while working alongside cross-functional teams to propel business outcomes and solve complex problems.

• Developed and executed computer vision algorithms for image processing and anomaly detection leveraging camera feeds.

• Implemented a Face AI application predicting wrinkles, dark circles, age, skin glow, and dark spots on human faces and achieved accuracy up to 93.50%.

• Deployed 7 web-based Artificial Intelligence and Machine Learning applications on the Heroku server.

• Led end-to-end deployment of deep learning models on cloud platforms, specifically on Azure and Google Cloud

• Produced reports with the help of SQL Server, MySQL, Talend ETL Tool, and created dashboards to visualize dataset for healthcare and manufacturing clients based on specific requirements. Improved customer feedback and suggestions.

• Proposed and contributed to two value-added projects, including a customer retention program in e-commerce and a web-based duplicate checker tool. Used Python for coding and worked on code review.

• Leveraged advanced machine learning and statistical techniques to design and evaluate algorithms for improving model performance, quality, and accuracy up to 96.25%.

• Utilized MLFlow for model serving and monitoring on Databricks, ensuring robust performance and reliability of machine learning models in production environments.

• Data Quality Assessment (Assessment of data quality and completeness in preparation for analysis) (i) Perform Data Quality Analysis and Analytical Dashboard Creation

• Data Insights (Targeting high-value customers based on customer demographics and attributes) (i) Perform Data Analytics and Customer Segmentation based on demographics

• Perform Data Visualisations and make Data Dashboards on Tableau for the presentations

• Implemented and optimized traditional machine learning algorithms (e.g., Random Forest, SVM, KNN) for image classification tasks, improving model accuracy by 15% and reducing processing time by 30%.

• Developed a computer vision pipeline using OpenCV and TensorFlow for real-time object detection, achieving 90% accuracy on a custom dataset of 10,000 images.

• Collaborated on feature engineering and selection processes, utilizing statistical techniques and domain knowledge to enhance model performance for predictive maintenance applications.

• Developed and implemented machine learning models using Python and scikit-learn, resulting in a 20% improvement in predictive accuracy for customer churn analysis.

• Collaborated with cross-functional teams to clean, preprocess, and analyze large datasets using SQL and Pandas, contributing to data-driven decision-making processes.

• Assisted in creating and maintaining interactive data visualizations using Tableau, effectively communicating complex insights to non-technical stakeholders.

• Conducted ETL operations to extract and transform market data from various sources, including LinkedIn, through iMacros and Python to support the LinkedIn campaigns.,

• Utilized Python and SQL (Relational Database) to analyze and clean data and worked alongside diverse teams to generate reports, including predictive analytics to forecast sales and enhance profitability for client’s e-commerce business.

• As a Student Ambassadors, I helped in spreading the word about Cognizance and expand the outreach to every region of the country.

• Member of the Student Partnership Programme (SPP) of IIT-Roorkee.

• Handled complex datasets using Python and learned to clean data by appropriate methods.

• Parallelly performed Join operations on multiple datasets to form the data into the desired format using SQL.

• As a Data Science Researcher at Galgotias University, I successfully authored and published two impactful research papers in the dynamic realms of data science and artificial intelligence.

• These contributions underscore my commitment to advancing knowledge in these fields and solidify my expertise as a researcher dedicated to innovation and excellence.