Vansh Bansal's Resume

Vansh Bansal

Ambitious Problem Solver with a flair for creating meaningful AI/ML applications and insights.

Delhi/DXB

About

Driven AI/ML Engineer committed to delivering innovative, scalable solutions that drive real-world impact. With a proven track record in designing end-to-end AI/ML systems and optimizing data pipelines, I bring a results-oriented approach to problem-solving. My ability to seamlessly integrate advanced AI technologies into business processes ensures measurable improvements and strategic advantage.

Work Experience

FinBox

Mar 2025 - Aug 2025

Data Science Intern

Gurgaon, India
  • Designed and deployed multiple LSTM-based NLP models to Amazon s3 to classify SMS messages across 5 distinct tagging dimensions, achieving 97-99% accuracy per tag and 95.5% overall multi-label accuracy, enhancing internal workflows.

  • Currently working on fine-tuning open-source LLMs such as Qwen, LLaMA, and Mistral to automate manual data enrichment tasks, with the goal of reducing workforce dependency by up to 60% and accelerating data enrichment pipelines.

  • Reduced latency of the DeviceConnect product by ~20% through codebase profiling and optimization using tools like cProfile, snakeviz, Cprogram, improving user responsiveness and lending journey.

Education

Vellore Institute of Technology

2021 - 2025
Bachelor's Degree in Information TechnologyVellore, Tamil Nadu

Skills

  • C/C++
  • Python
  • HTML
  • SQL
  • NumPy
  • Pandas
  • Matplotlib
  • BeautifulSoup
  • Selenium
  • Tableau
  • scikit-learn
  • LangChain
  • Git
  • Data Structures and Algorithms
  • Data Analysis
  • Machine Learning
  • Deep Learning
  • Generative AI
  • LLMs
  • OpenAI
  • HuggingFace
  • Unsloth

Projects

Network Security Data - End-to-End ML Project

End-to-end ML project with a multi-stage pipeline, model tracking via MLFlow, and containerized deployment using Docker for real-time network security predictions.

  • MongoDB
  • MLFlow
  • Dagshub
  • FastAPI
  • Docker
  • Python

Social Media Analytics for Influencer Insights

Analytics system to identify top influencers by extracting faces from videos, calculating performance scores based on engagement metrics, and delivering actionable insights through a 3-stage pipeline.

  • Python
  • Pandas
  • DeepFace
  • Flask

Dynamic AI Character: Social Media Post Generator

AI character simulating human-like behavior to generate diverse social media posts in text, image, audio, and video formats, with seamless Twitter integration for direct posting.

  • Python
  • Langchain
  • Langgraph
  • OpenAI
  • Tweepy

Generative AI for Automated Resume Screening

Generative AI solution automating resume parsing, scoring, and batch processing to streamline the recruitment process, with GPT integration for data extraction and contextual relevance.

  • Python
  • Langchain
  • Langgraph
  • OpenAI
  • Pandas

Comprehensive Credit Score Analysis and Predictive Modeling

In-depth credit score analysis with predictive modeling, featuring data cleaning, visualizations, and dimensionality reduction to achieve 76% accuracy in credit score predictions.

  • Sklearn
  • Pandas
  • Numpy
  • Matplotlib

Energy Price Prediction Using ML

Machine learning model to predict next-day energy prices, enhancing accuracy by 45% with advanced techniques like PCA, GridSearchCV, and comprehensive data pipeline engineering.

  • Sklearn
  • Pandas
  • Numpy
  • Matplotlib
  • Seaborn

Indian Startup Ecosystem Analysis

Analysis of trends in the Indian startup ecosystem using funding data, covering funding trends, geographical distribution, investment types, top industries, and key players.

  • Numpy
  • Matplotlib
  • Pandas

IMDB Movie Dataset Analysis

In-depth analysis of the IMDB movie dataset using SQL queries and pandas, exploring earnings, ratings, runtime, budget, and genre statistics.

  • SQL
  • Pandas
  • Python