I am a specialized professional in the AI/ML and Data Engineering space, pursuing my MS in Information Systems with a Data and ML/AI specialization at Northeastern University (GPA: 3.95/4.0). My expertise is concentrated in designing and implementing:
- Advanced RAG Systems: Developed semantic chunking architectures achieving 92% retrieval accuracy with context-aware vector storage optimizations
- Multi-Agent Orchestration Frameworks: Built specialized state-based agent networks using LangGraph for complex task coordination
- Efficient ML Model Fine-tuning: Implemented parameter-efficient techniques (LoRA, adapters) reducing training resources while improving model performance
- Hybrid Neural Architectures: Created combined CNN-RNN-LSTM models for multi-modal data integration with optimized prediction capabilities
- Scalable Data Pipelines: Designed production-grade ETL workflows with Apache Airflow for high-throughput document processing
My technical approach fuses cloud-native infrastructure with specialized AI components, consistently delivering production-ready systems that combine theoretical understanding with practical implementation expertise.
Machine Learning & AI
Transfer Learning, Reinforcement Learning, Ensemble Methods, Feature Engineering, PEFT, Weights & Biases, Hyperparameter Optimization, Model Quantization, Federated Learning, AutoML, Model Explainability, Bayesian Optimization, Data Augmentation
Natural Language Processing
RAG Systems, Vector Databases, Prompt Engineering, Multi-Agent Systems, Text Embeddings, Semantic Search, LLMs, NLTK, SpaCy, Named Entity Recognition, Sentiment Analysis, Language Generation, Topic Modeling, Word Embeddings, Tokenization
Data Engineering & Big Data
ETL/ELT Pipelines, Data Warehousing, Data Lakes, Stream Processing, Batch Processing, Data Modeling, Dimensional Modeling, Data Governance, Dask, PySpark, Hadoop, Data Validation
Cloud Computing & MLOps
CI/CD, GitHub Actions, MLflow, Model Deployment, Infrastructure as Code, Monitoring, Logging, Serverless, Containerization, Microservices Architecture, Ansible, Jenkins, Cloud Security, Cost Optimization, Hybrid Cloud
Deep Learning & Neural Networks
Attention Mechanisms, Transfer Learning, Autoencoders, Generative Models, Diffusion Models, Neural Architecture Search, GANs, Graph Neural Networks, Variational Autoencoders, Optimizers, Loss Functions, Batch Normalization, Dropout, Activation Functions
Computer Vision
Image Classification, Object Detection, Image Segmentation, Face Recognition, Image Processing, Feature Extraction, OCR, Transfer Learning with Visual Models, Image Generation, Video Analysis, Depth Estimation, Image Enhancement, Style Transfer, Pose Estimation
Software Development
OOP, Design Patterns, API Development, Git, GitHub, CI/CD, Testing, Documentation, Code Review, Shell Scripting, REST API, Microservices, Web Development, Next.js, React, TypeScript
Database Management
SQL, NoSQL, Database Design, Query Optimization, Indexing, Database Migrations, Vector Databases, MySQL, Snowflake, DynamoDB, Time-series Databases, Graph Databases, Stored Procedures
Data Analysis & Visualization
Exploratory Data Analysis, Statistical Analysis, Data Cleaning, Feature Engineering, Dashboard Creation, Data Storytelling, Power BI, Tableau, Interactive Visualizations, Time Series Analysis, Correlation Analysis, Dimensionality Reduction, Clustering, DBT
Transform static PDFs into dynamic, personalized learning experiences through advanced RAG and multi-agent intelligence
Key Achievements:
- Engineered semantic chunking with dynamic thresholding achieving 92% retrieval accuracy
- Orchestrated five specialized agents generating diverse learning formats (podcasts, flashcards, quizzes)
- Reduced API costs by 41% through Upstash semantic caching with 97% similarity threshold
Technologies: LangGraph, OpenAI, Pinecone, FastAPI, Next.js, Airflow, AWS/GCP
Intelligent research system leveraging specialized agents for document analysis, web search, and academic paper retrieval
Key Achievements:
- Reduced research time by 65% while maintaining 95% semantic relevance
- Engineered document pipeline processing text, tables, and images for 4,500+ vectorized elements
- Implemented production-grade system with real-time streaming and JWT authentication
Technologies: LangGraph, OpenAI, Pinecone, Airflow, FastAPI, Docker, Streamlit
Advanced implementation of LoRA fine-tuning to adapt large language models for specialized SQL generation
Key Achievements:
- Achieved 86% improvement in BLEU score with rank-16 LoRA configuration
- Optimized Google's Gemma-3-1b-it model for natural language to SQL translation
- Implemented efficient inference pipeline with comprehensive evaluation metrics
Technologies: PEFT, HuggingFace, Google Gemma, PyTorch, Transformers
Targeted adaptation of Stable Diffusion for enhanced photorealism through U-Net component fine-tuning
Key Achievements:
- Focused fine-tuning strategy preserving CLIP and VAE while adapting U-Net for style transfer
- Implemented memory-efficient training techniques for consumer hardware compatibility
- Developed full-stack application with FastAPI/Streamlit for comparative analysis
Technologies: PyTorch, Diffusers, FastAPI, Streamlit, DDPM
Educational chatbot providing expert guidance on large language model concepts and implementation
Key Achievements:
- Created comprehensive knowledge base covering LLM fundamentals, RAG, fine-tuning, and deployment
- Implemented user profile management for personalized interaction histories
- Designed responsive UI with real-time streaming responses for enhanced UX
Technologies: LangChain, OpenAI API, Streamlit, Pydantic
Multi-modal precipitation prediction system integrating satellite imagery with meteorological data
Key Achievements:
- Engineered hybrid architecture combining CNN for spatial imagery and RNN-LSTM for temporal data
- Implemented advanced preprocessing for wind vector creation and cyclic date-time features
- Addressed class imbalance challenges in precipitation frequency distribution
Technologies: TensorFlow, Conv3D, LSTM, Batch Normalization, Dropout
High-performance classification system for 958K asteroid records using distributed computing techniques
Key Achievements:
- Reduced data preprocessing and model training times by 8.24x through parallel processing
- Optimized multi-CPU and GPU resource utilization on high-performance computing infrastructure
- Implemented and compared multiple ML algorithms (Random Forest, KNN, XGBoost, Neural Networks)
Technologies: Dask, PyTorch, Multiprocessing, Random Forest, XGBoost, Neural Networks
Multi-agent system for automated book writing with integrated research and content generation
Key Achievements:
- Orchestrated specialized AI agents for research, writing, and content organization
- Integrated web scraping capabilities for automated data collection and analysis
- Implemented end-to-end workflow from topic exploration to final content production
Technologies: CrewAI, Google's Gemma 3, BrightData, Ollama
Transformer-based model for low-resource language translation with advanced decoding strategies
Key Achievements:
- Implemented multi-head attention and positional encoding for context-aware translation
- Compared greedy search vs. beam search decoding strategies for optimal results
- Designed custom training regime with dynamic learning rate scheduling
Technologies: PyTorch, Transformers, Adam optimizer, Beam Search
Advanced image description system combining computer vision and natural language generation
Key Achievements:
- Developed attention mechanism focusing on relevant image regions during caption generation
- Implemented ResNet50-based feature extraction combined with GRU for sequential text generation
- Created flexible inference pipeline supporting both batch processing and URL-based image captioning
Technologies: TensorFlow, ResNet50, GRU, Bahdanau Attention, COCO dataset
Northeastern University, Boston | CGPA: 3.96/4.0 | May 2025
National Institute of Technology, Andhra Pradesh, India | June 2021
I'm always excited to collaborate on projects involving AI, machine learning, and data engineering. Whether you have a project idea, question about my work, or just want to connect, feel free to reach out!