Divyansh Jain
IIT Bombay graduate with experience in Machine Learning, Deep Learning, and Generative AI. Built multimodal AI systems, LLM-powered applications, and Retrieval-Augmented Generation (RAG) pipelines using PyTorch and LangChain through internships and academic projects. Seeking full time opportunities to develop scalable AI solutions for real-world applications.
Experience
AI Engineer Intern
ReachIvy
Mumbai
Built a multimodal ML pipeline integrating Whisper (speech), OpenCV (facial analysis), and MediaPipe (gesture tracking) to automate candidate interview assessment, reducing manual review effort by 60%
Developed an embedding-based response-grading module using SentenceTransformers (all-MiniLM-L6-v2) and cosine similarity, enabling real-time automated interview scoring across 100+ candidate evaluations
Automated the end-to-end interview review workflow, reducing per-candidate evaluation time by 50% while standardizing the intern hiring process across hiring teams
Education
Indian Institute of Technology Bombay
Bachelor of Technology
CGPA: 7.6/10
Relevant Coursework: Machine Learning, AI & Data Science, Digital Image Processing, Optimization,
Probability & Statistics
Skills
Projects
AI Resume Parser
LangChain, LLMs, RAG, YouTube Data API
Built LangChain base RAG resume analyzer with 6 chunking strategies and 5 embedding models
Designed factory-pattern abstraction for 4 LLM providers with runtime switching via REST APIs
Engineered ATS skill-gap engine scoring resumes against job descriptions and mapping gaps to top-3 YouTube
API course recommendations per skill
MRI-Based Brain Tumor Classification
Vision Transformer, PyTorch, Fine Tuning
Fine-tuned a Vision Transformer (PyTorch) for 4-class tumor classification with a custom DataLoader pipeline
Achieved 93% test accuracy using L2 regularization and deployed the model via a Gradio inference interface
Content-Based Movie Recommender
BERT Embeddings, Similarity Search, Streamlit
Built a content-based recommender using BERT embeddings and cosine similarity over 44k movie descriptions
Engineered 5 recommendation modes ranking by overview, genre, production house, keywords, and cast
Deployed a Streamlit app with movie detail views and a paginated catalog spanning 480 pages of titles
Rental Price Prediction
Regression, XGBoost, TF-IDF, EDA
Performed EDA, feature engineering, and data cleaning on 74K rental listings, benchmarked 4 regression
models
Engineered TF-IDF text features, target encoding and applied stacked regression ensemble to achieve R
2
= 0.83
Honors and Achievements
JEE Advanced 2022
Secured All India Rank 2416 among 170,000+ candidates
JEE Main 2022
Achieved 99.4 percentile among 1.2 million candidates nationwide
CBSE Class 12 Board Examinations
Scored 97% overall with a perfect 100/100 in Mathematics