I BUILD AI SYSTEMS. I SHIP CODE.
Hey, I'm Dhruv Garg — Software Engineer & AI/ML Developer. I build RAG pipelines, deep learning models, and intelligent systems that solve real-world problems. Currently interning at Blostem, processing 10K+ documents with under 2s retrieval latency.
CS STUDENT TURNED AI ENGINEER
I'm a Computer Science student at BML Munjal University with a passion for building intelligent systems that make a real impact. From designing end-to-end RAG pipelines processing thousands of documents to training deep learning models with 95%+ accuracy, I thrive at the intersection of software engineering and AI.
Currently working as a Software Engineer Intern at Blostem, where I'm building production-grade AI systems. I believe in writing clean code, shipping fast, and letting data drive decisions — whether it's optimizing retrieval latency or building interactive dashboards.
"The best way to predict the future is to build it — one model at a time."
THE TECH STACK I WIELD
Languages
Frameworks
Dev Tools
AI/ML Libraries
GenAI & RAG
REAL-WORLD EXPERIENCE
Software Engineer Intern
- Designing and developing a full end-to-end RAG pipeline processing 10K+ documents with under 2s average retrieval latency.
- Implementing document ingestion, chunking, and embedding workflows using LangChain and LlamaIndex, reducing retrieval noise by 30%.
- Integrating Hugging Face models and OpenAI API, improving response relevance by iterating on 15+ prompt templates.
Data Analyst Intern
- Cleaned and preprocessed 20K+ rows of business data using Pandas, resolving missing values, outliers, and format inconsistencies.
- Built 5+ interactive dashboards using Matplotlib to surface key business trends across 3 business verticals.
- Automated repetitive data cleaning workflows via Python scripts, cutting manual preprocessing time by 40%.
FROM IDEA TO DEPLOYED SOLUTION
AI vs Real Image Detector
Deep-learning image forensic classifier using EfficientNetV2-B0 combined with FFT frequency-domain analysis to distinguish AI-generated images from real photographs. Deployed as an interactive Streamlit app with Grad-CAM visual explanations.
AI Interview Analysis
End-to-end AI platform extracting 15+ behavioral metrics including gaze tracking, posture detection, and facial emotion recognition from interview videos. Features a RAG pipeline using LangChain and ChromaDB for structured feedback generation.
Flight Crash Analysis
Analyzed 50K+ aviation incident records across 70+ years to identify crash risk factors, seasonal patterns, and operator-level trends. Trained 5 ML models including Random Forest and XGBoost, achieving 89% prediction accuracy.
WHERE IT ALL STARTED
BML Munjal University
READY TO COLLABORATE?
Whether it's an exciting AI project, an internship opportunity, or just a tech conversation — I'd love to hear from you.