Simran Kathalkar

AI/ML Engineer with 4+ years of enterprise experience building LLM-driven, Generative AI, and Agentic AI systems, including recommendation engines and cosine similarity–based intelligence. Expert in Python, FastAPI, databases, prompt engineering, and deploying scalable AI solutions using AWS, Azure DevOps, Azure Foundry, Jenkins, GitHub, and GitLab.

• AZ-900: certified

• PROFESIONAL EXPERIENCE:

► Technical Engineer I Kyndryl, Pune 3.1 Years I September 2021 to Present Date

• PROJECTS:

Project Name: Sentiment analysis using Natural language processing.

Domain: IT Data and Analytics

• Description:

A sentiment based binary review model and labelling on fresh reviews1

• Responsibilities:

• Data Categorization and Restructuring: To extract text data from client-provided CSV and image files using OCR and Python modules like Pandas.

• Execute text processing techniques like stemming and lemmatization and stop word elimination libraries reducing preprocessing time per document by 50% libraries.

• To perform Tokenization and create dataframe by using Count vectorizer like Bag of words, TF-IDF, Word2Vec and Doc2Vec.

• Feature extraction and feature selection based on Inter class and Intra class feature weightage.

• To work in coordination with various functional teams to build models and track results.

• To track and analyze model performance and data accuracy, develop procedures and tools.

Project Name: Employee Turnover prediction using Machine Learning

Domain: IT Data and Analytics

• Description:

Create a system of prediction of employee attrition and retention by visualization using business analytics principles using ML algorithms. Use predictive analysis to improve the working environment for employees.

• Responsibilities:

• Predictive analysis is used to calculate how many employees a business will require if certain workforce quit.

• Visualized data to gain understanding of it and identified data distribution that affects performance of model.

• Verified data quality via data cleaning and introduced validation strategies.

• Implemented preprocessing and feature engineering to be done on given dataset.

• Prepared data augmentation pipelines and trained models.

• Analyzed the errors and designed strategies like tuning hyperparameters of model to improve model performance.

• To resolve the main causes of the high turnover percentage.

• To Understand general patterns, examples, and manifestations of wearing down so explicit activities plans can be set up for each pattern, example, and side effects.

Experience at Accenture: 1 year 3 months, currently working

Project 1: Client : TIAA

• Built a machine learning–based recommendation engine for TIAA using cosine similarity and Python to generate accurate, data-driven recommendations.
• Developed ML pipelines and integrated MongoDB for scalable data storage and retrieval.
• Delivered an end-to-end AI solution translating complex data into meaningful business recommendations.

Project 2: Client : Baker Hughes, currently working:

Designed and delivered a multi-agent Generative AI solution leveraging Agentic AI architecture to automate complex workflows and decision-making processes for Baker Hughes.
Implemented LLM-powered agents using LangChain and LangGraph, enabling agent collaboration, task orchestration, and context-aware reasoning.
Built embedding-based retrieval and similarity pipelines to enhance response accuracy and contextual understanding across agents.
Integrated Azure Foundry for model management and orchestration, and Azure DevOps for CI/CD, version control, and automated deployments.
Utilized Azure Cosmos DB for scalable storage of conversation state, embeddings, and agent memory.
Collaborated with stakeholders to translate business requirements into production-ready GenAI solutions, ensuring scalability, security, and enterprise compliance.Tech Stack: Generative AI, Agentic AI, Multi-Agent Systems, LangChain, LangGraph, Embeddings, Azure Foundry, Azure DevOps, Cosmos DB.
Built a multi-agent Generative AI solution for Baker Hughes using Agentic AI, LangChain, and LangGraph, enabling intelligent task orchestration and decision automation.
Implemented embedding-based retrieval and integrated Cosmos DB, with end-to-end deployment using Azure Foundry and Azure DevOps.

• English
• Marathi
• Hindi