Bachelor of Engineering (BEng)

During my internship at Splashgain, I was responsible for cleaning and maintaining educational data in the company's tool. I interacted with both domestic and international users to resolve any issues with the product and developed a passion for data.

Data Cleaning:
My role involved ensuring the accuracy and quality of the educational data in the company's tool through effective cleaning methods. This required attention to detail and an understanding of data management best practices.

Interaction with Users:
I was the point of contact for users and was responsible for resolving any issues they faced with the product. I communicated with users from both domestic and international locations to provide them with timely and effective solutions.

Developed Passion for Data:
Through my work with educational data and interaction with users, I developed a strong interest in data and its potential for improving products and processes. This experience further fueled my passion for data and its

Associate

- Created models using Keras CNN and Huggingface ViT Transformer for table classification and border identification. Also, custom models for document categorization.- Developed models for predicting column lines using Detectron2 and annotation tools.- Contributed to an NLP model using spaCy to extract specific elements from drug names.- Established an automated pipeline for processing 50+ healthcare/insurance templates. Utilized Camelot and Tabula for table data extraction, managed code with Git, and addressed client-requested changes

Junior Data Scientist

<ul><li>Led end-to-end development of a hybrid query routing POC (LangChain, Qwen, ChromaDB) for a major auto insurance client, securing project win. </li><li>Engineered a dynamic SQL query pipeline (LangChain, Qwen/Groq, Postgres) for accurate LLM-generated responses. </li><li>Built a RAG workflow (ChromaDB, Google Text Embedding 004, UnstructuredLoader) for context-rich policy question answering.</li><li>Developed a GenAI LLM module (LangChain, Llama 3.1/Ollama) to automate resume extraction, reducing HR lookup time by 80%. </li><li>Built a Top 5 Skills Identification Module (Llama 3.1) increasing skill categorization accuracy by 89% over manual methods. </li><li>Led architectural decisions for LLM-related modules (resume extraction, skill identification) using LangChain.</li></ul>

Data Scientist

I write, build, and explore around AI, data, and technology.

projects

hashnode

github

Data Scientist | Experience in Model Development Lifecycle | Building Innovative Data-Driven Solutions🚀

Shreyas Kulkarni

I write, build, and explore around AI, data, and technology.