ANKIT BISWAS

• Collaborate with cross functional Applied AI team and build statistical / predictive ML models
• Develop data ingestion process to automate data transfer from multiple internal sources into single database by creating workflows through Alteryx
• Create Alteryx Macros to improve existing workflows, reduce repetitive tasks, and increase re-usability
• Improve data quality using several transformation techniques and regular expressions to align data with expectations of the business
• Created data models and documented data flows to improve system architecture and data quality
• Create Python scripts to ingest data from external vendors through API calls and then parse, transform, and clean data before storing it in database.

• Assist in developing, implementing, and deploying data science models and algorithms to solve complex business problems
• Collaborate with cross-functional teams, including data engineers, business analysts, and consultants, to gather and understand project requirements
• Build predictive and prescriptive models using machine learning and statistical techniques, leveraging tools such as Python, R, or SQL
• Assist in analyzing and interpreting model results, providing insights and recommendations to stakeholders
• Support development of data visualization dashboards and reports to communicate findings effectively
• Contribute to data exploration and preparation, performing data cleaning, transformation, and feature engineering tasks to ensure data quality and relevance.
• Increased customer satisfaction by resolving complex issues and providing exceptional service at all times.
• Established strong relationships with key stakeholders, facilitating effective communication across departments and teams.
• Played an integral role in launching successful projects by coordinating tasks among team members while adhering closely to established timelines and budgets.

• Access and compile data sets from various sources for exploratory and pre-defined analyses
• Audit data sets for completeness, validity, and other pertinent data health measures
• Transform data for specific team functions or processes as appropriate
• Conduct statistical testing and exploratory data analysis
• Create basic data visualizations.

• Responsibilities Requirement gathering, Use SQL to query databases and generate data for analysis and insights based on use cases and requirements
• Requirement analysis, System Design, Writing test cases, Perform Unit and Integration testing.
• Improved system performance by identifying and resolving technical issues promptly.
• Analyzed system requirements, developing tailored solutions that met or exceeded client expectations.

KYC Automation Project – Data Integration and ETL Development

• As part of a broader KYC automation project, we aimed to reduce manual efforts by automating over 30 key fields in a form with 100+ inputs. We sourced data from a third-party vendor (BVD) but faced compliance challenges with direct API access. Upon discovering the data was already stored within another team's platform, we accessed it via their internal API. Using Alteryx, we developed an ETL pipeline to clean, transform, and store the data in a tabular format. This involved overcoming challenges with JSON data by employing advanced regex techniques to ensure the data met specific quality and formatting requirements.

LLM Prompt Engineering for KYC Process Optimization

• In a KYC-related LLM Prompt Engineering project, we aimed to fast-track tasks like comparing related parties and beneficial owners across lifecycle documents. Initial single-step prompts yielded inconsistent results. Through iterative optimization, we improved outcomes by breaking the process into step-by-step prompts, incorporating outlier scenarios and key business insights. After each step, we instructed the LLM to review and refine its responses, treating it as a "student" learning from feedback to generate more accurate answers. This approach consistently delivered refined outputs, effectively addressing both straightforward and complex tasks requiring broader business understanding and validation against guidance documents.

Source Code Analysis Gen AI

• Developed an interactive Q&A system to simplify understanding large codebases. The project clones a GitHub repository, processes Python files into structured chunks using LangChain, creates a searchable knowledge base with OpenAI embeddings stored in a Chroma vector database, and integrates a conversational interface using OpenAI's ChatGPT and memory for context-aware responses. The system enables seamless querying of code insights, like understanding classes or methods, enhancing developer productivity.
• Tech Stack: Python, Git, LangChain, OpenAI API, Chroma. Highlights: Automated knowledge extraction, semantic search, and conversational AI for efficient codebase exploration.

Customer Segmentation Model for a US Retail Giant

• Performed movement analysis utilizing RFM on various customer segments to analyze their evolution over multiple historical time
• frames.
• Developed a customer segmentation machine learning model using an RFM approach for the retail client's marketing team, enabling optimization of customer retention and acquisition strategies through predicting future customer segments.

Market Basket Clustering for a US Retail Giant

• Conducted unsupervised machine learning analysis on spending
• patterns for a major retail player, aimed at deriving distinct customer groups and informing targeted marketing strategies. Leveraged spending proportions of categorical and non-categorical
• features to construct the features. Employed K-means modeling to unveil purchasing preferences and product associations within these clusters.
• Transformed cluster insights into compelling narratives to guide strategic marketing approaches.

Employee Attrition Prediction Analysis for HR

• This was a data analytics project in which we prepared a model for the organization to predict Attrition and provide insights from the data about the important factors associated with it so that the organization can take corrective or preventive measures to stop or control it.

More projects - https://github.com/ankitbiswas/GenAI.git