AI & Machine Learning Engineer — B.Sc Hons Computer Science, Hansraj College, University of Delhi
I'm an AI/ML engineer building production-grade systems across the full stack — from LLM pipelines and agentic architectures to MLOps, big data, and cloud deployment. I focus on systems that are functional, well-engineered, and deployable.
Currently building LexShield AI, a multi-agent legal intelligence platform using LangGraph, 6-layer memory architecture, hybrid RAG, and LLM-based intent classification — designed to help citizens understand their legal rights in India.
- 🔭 Focus areas: Agentic AI · RAG Systems · MLOps · AIOps · Deep Learning · Big Data Engineering
- 🛠️ Serving: All major projects are production-served via FastAPI, Streamlit, or Docker
- 💬 Ask me about: LLMs, RAG, LangGraph agents, LoRA fine-tuning, MLflow, Spark, or deploying ML at scale
| Project | Description | Key Tech |
|---|---|---|
| ⚖️ LexShield AI | AI-powered Indian legal intelligence platform. Multi-agent LangGraph system with hybrid BM25+semantic RAG, 6-layer memory architecture, LLM-based intent classification, multilingual OCR (English/Malayalam/Hindi), spaCy NER, and clause-level risk scoring over IPC, BNS, CrPC, and Kerala statutes. | LangGraph · ChromaDB · Groq LLaMA · spaCy · PyMuPDF · FastAPI |
| 🏷️ Multi-Label Text Classification | Production-grade multi-label text classification pipeline. DistilBERT fine-tuned with MLflow experiment tracking, FastAPI serving, and full Docker deployment. | DistilBERT · MLflow · FastAPI · Docker |
| 🌿 LoRA Image Classifier | LoRA fine-tuned Vision Transformer for plant disease classification with Grad-CAM explainability. Demonstrates parameter-efficient fine-tuning on a constrained resource budget. | ViT · LoRA · Grad-CAM · FastAPI |
| 🔍 AIOps Log Anomaly Detection | Real-time AIOps monitoring engine using Isolation Forest to detect structural anomalies in unstructured server logs. Dual interface via FastAPI and Streamlit. | Isolation Forest · FastAPI · Streamlit |
| 📈 LSTM Anomaly Detection System | Multivariate time series anomaly detection using an LSTM Autoencoder. Full MLOps setup with MLflow, Plotly Dash dashboard, FastAPI, and Docker. | LSTM · MLflow · Plotly Dash · FastAPI · Docker |
| 🚕 Spark ETL + ML Pipeline | End-to-end Apache Spark ETL and ML pipeline on 9.3M NYC taxi trips. Includes Streamlit dashboard, FastAPI endpoint, and Docker deployment. | Apache Spark · PySpark · Streamlit · FastAPI · Docker |
| 🏭 Retail Log Hadoop Pipeline | Production-inspired distributed data pipeline using HDFS, MapReduce Streaming, and Hive for retail log analytics. | Hadoop · HDFS · MapReduce · Hive |
| 👥 Customer Retention & CLV Engine | End-to-end ML pipeline predicting churn and Customer Lifetime Value. Served via FastAPI and Streamlit. | Scikit-learn · FastAPI · Streamlit |
| ☁️ AuthAPI — Multi-Cloud Deployment | Authentication API deployed across Heroku, AWS EC2, and AWS Elastic Beanstalk. Demonstrates multi-cloud deployment and CI/CD configuration. | Python · REST API · Heroku · AWS |
| 📊 Data Science Projects | Collection of end-to-end ML experiments covering regression, EDA, and predictive modelling on real-world datasets. | Python · Scikit-learn · Pandas · Jupyter |
"Build systems that reason, not just predict."