AI Engineer at Zasti Inc | MS in Computer Science from University of Cincinnati (GPA: 3.95/4.0) | Specializing in Agentic AI, Multi-Modal RAG Systems, and LLM Optimization

AI Specialist with hands-on expertise in agentic AI, multi-modal RAG systems, and LLM optimization for impactful healthcare and research solutions. Passionate about designing ethical and reliable AI tools that automate complex workflows, boost model accuracy, and empower learning for technical and non-technical audiences alike. My work spans from cutting-edge research in neuroimaging to production-ready AI systems transforming healthcare and clinical research.

Master of Science in Computer Science
University of Cincinnati, Cincinnati, OH | August 2022 - May 2024

• GPA: 3.95/4.0
• Thesis: "A Multimodal Neuroimaging Method for the Prediction of Visual Stimuli: A Temporal Convolutional Network-Based Approach for EEG-fMRI Fusion"
• Advisor: Prof. Vikram Ravindra
• Committee: Prof. Jun Bai, Prof. Vesna Novak
• Focus: Multimodal Deep Learning, Neuroimaging, Brain-Computer Interfaces, Temporal Convolutional Networks
• Award: Graduate Incentive Award for Academic Excellence ($60,000 scholarship)

MicroMasters in Data Science
University of California San Diego, San Diego, CA | April 2021 - January 2022

• Specialized in advanced data science, machine learning, and statistical methods
• Focus on practical applications of data science in real-world scenarios

Bachelor of Technology in Computer Science
SRM University, India | May 2018 - April 2022

• GPA: 3.98/4.0
• Focus: Machine Learning, Deep Learning, Data Structures, Algorithms
• Strong foundation in computer science fundamentals and software engineering

May 2024 - Present | Ashburn, VA

Architecting and deploying cutting-edge AI systems for healthcare and clinical research:

Agentic AI & Multi-Agent Systems:

• 🤖 Automated 95% of clinical research workflows by deploying agentic AI systems with multi-agent orchestration, streamlining task execution and collaboration across research teams
• 🔄 Designed and implemented autonomous agent frameworks for complex research workflows, enabling intelligent task decomposition and execution

Advanced RAG Systems:

• 🚀 Architected Raptor RAG system with hybrid knowledge graph integration and evidence database injection, achieving 45% faster query processing and 38% improved factual accuracy across 50M+ document embeddings
• 📊 Engineered multi-modal RAG pipeline with web scraping capabilities and hallucination detection for medical images and text, automating clinical workflows across 15+ lab report formats

LLM Optimization & Deployment:

• ⚡ Optimized LLM deployment through GPTQ 4-bit quantization, reducing model size by 75% and inference costs by $2K monthly while maintaining 98% performance parity
• 🎯 Implemented advanced prompt engineering and fine-tuning strategies for domain-specific healthcare applications

Leadership & Education:

• 👨‍🏫 Trained five interns and led three workshops in AI, ML, and agent development, designing customized educational materials with adaptive AI agents for tailored learning
• 📚 Created comprehensive training programs bridging technical complexity with accessible learning

Technologies: Python, LangChain, LlamaIndex, PyTorch, Hugging Face Transformers, Vector Databases (Pinecone, Chroma), GPTQ, RAG Systems, Knowledge Graphs, Multi-Agent Orchestration

August 2022 - April 2024 | Cincinnati, OH

Empowering the next generation of AI practitioners through hands-on education:

Teaching & Mentorship:

• 👥 Led instruction and mentorship for 500+ students by designing and delivering hands-on workshops in Python, cloud computing, and machine learning, achieving 90% assignment completion rate
• 📈 Increased student engagement by 40% through development of scalable, user-friendly educational resources including interactive coding tutorials, project guides, and assignment walkthroughs

Curriculum Development:

• 🎓 Developed and launched interactive coding tutorials and project guides for AI/ML courses, empowering students to deploy real-world ML models and cloud-based solutions
• ✨ Promoted ethical and responsible AI practices, demonstrated by positive feedback from 95% of participants
• 🔧 Collaborated with faculty to create accessible resources that made advanced AI/ML techniques available to learners of all backgrounds

Impact:

• Significant boost in student competency in AI concepts and practical implementation
• Successfully bridged theory and practice, enabling students to build production-ready solutions

February 2021 - July 2022 | Remote

Led development of deep learning solutions for medical applications:

Technical Leadership:

• 🏆 Led a 4-member team to create deep learning algorithms for medical product, boosting system efficiency by 45%
• 🎯 Designed innovative pattern recognition algorithm that enhanced predictive accuracy by 25%, accelerating clinical decision support capabilities

Product Development:

• 📊 Pioneered interactive Power BI dashboards to track algorithm performance, leading to 10% improvement in diagnostic accuracy through data-informed decision making
• 🔬 Demonstrated strong technical leadership in model development and deployment for healthcare applications

Technologies: TensorFlow, PyTorch, Computer Vision, Pattern Recognition, Power BI, Medical Imaging

My research sits at the intersection of neuroscience, machine learning, and artificial intelligence:

• 🧠 Multimodal Learning & Fusion - Integrating heterogeneous data sources (EEG, fMRI, vision, language)
• 🤖 Agentic AI - Autonomous agents, multi-agent systems, tool use, and reasoning
• ✨ Generative AI - Diffusion models, LLMs, vision-language models, foundation models
• 🏥 Medical AI & Neuroimaging - Brain-computer interfaces, neural decoding, clinical applications
• 🔄 Continual Learning - Lifelong learning, catastrophic forgetting, parameter-efficient fine-tuning
• 🎯 Attention Mechanisms - Transformers, cross-modal attention, token-level representations

• Healthcare & Medical Imaging
• Brain-Computer Interfaces
• Climate Tech & Sustainability
• Autonomous Systems
• Human-AI Collaboration

Python ████████████████████ Expert
MATLAB ████████████████░░░░ Advanced
R      ████████████░░░░░░░░ Intermediate
SQL    ████████████░░░░░░░░ Intermediate

Frameworks & Libraries:

• Deep Learning: PyTorch, TensorFlow, Keras, JAX
• Transformers: Hugging Face Transformers, LangChain, LlamaIndex
• Computer Vision: OpenCV, torchvision, Albumentations, timm
• NLP: spaCy, NLTK, Gensim, sentence-transformers
• Scientific Computing: NumPy, SciPy, scikit-learn, Pandas

Specialized Libraries:

• Neuroimaging: MNE-Python, Nilearn, NiBabel, EEGLAB, SPM
• Generative AI: Stable Diffusion, DALL-E, GPT APIs, Anthropic Claude
• Agentic AI: LangChain, AutoGPT, LangGraph, CrewAI
• Model Optimization: ONNX, TensorRT, Quantization, Pruning

Core Machine Learning:

• Supervised Learning (Classification, Regression)
• Unsupervised Learning (Clustering, Dimensionality Reduction)
• Semi-supervised Learning
• Reinforcement Learning (Q-Learning, Policy Gradients)
• Transfer Learning & Domain Adaptation

Deep Learning Architectures:

• Convolutional Neural Networks (CNNs) - ResNet, EfficientNet, Vision Transformers
• Recurrent Neural Networks (RNNs) - LSTM, GRU, Bidirectional architectures
• Temporal Convolutional Networks (TCNs) - Causal convolutions, dilated convolutions
• Transformers - BERT, GPT, Vision Transformers (ViT), CLIP
• Attention Mechanisms - Self-attention, cross-attention, multi-head attention
• Graph Neural Networks (GNNs) - GCN, GAT, GraphSAGE

Generative AI:

• Large Language Models (LLMs) - GPT-4, Claude, Llama, Mistral
• Diffusion Models - Stable Diffusion, DALL-E, Imagen
• Vision-Language Models - CLIP, BLIP, Flamingo
• Variational Autoencoders (VAEs) - β-VAE, VQ-VAE
• Generative Adversarial Networks (GANs) - StyleGAN, CycleGAN, Pix2Pix

Agentic AI & Advanced Systems:

• Multi-Agent Systems - Coordination, communication, emergent behavior
• Tool Use & Function Calling - Agent-tool interaction, API integration
• Planning & Reasoning - Chain-of-thought, tree-of-thoughts, ReAct
• Memory Systems - Vector databases, episodic memory, long-term memory
• Autonomous Decision Making - Goal-oriented behavior, task decomposition

Parameter-Efficient Fine-Tuning (PEFT):

• LoRA (Low-Rank Adaptation)
• QLoRA (Quantized LoRA)
• Adapters (Houlsby, Pfeiffer)
• Prefix Tuning
• Prompt Tuning
• P-Tuning v2

Multimodal Learning:

• Early Fusion, Late Fusion, Hybrid Fusion
• Cross-Modal Attention
• Multimodal Transformers
• Vision-Language Pre-training
• Audio-Visual Learning
• EEG-fMRI Integration

Medical Imaging & Neuroimaging:

• Medical Image Segmentation (U-Net, nnU-Net)
• Image Registration & Alignment
• Brain Signal Processing (EEG, fMRI, MEG)
• Feature Extraction & Engineering
• Clinical Decision Support Systems

• Experiment Tracking: Weights & Biases, MLflow, TensorBoard
• Model Deployment: FastAPI, Flask, Docker, Kubernetes
• Cloud Platforms: AWS (SageMaker, EC2, S3), Google Cloud (Vertex AI), Azure
• Distributed Training: Ray, Horovod, DeepSpeed
• Version Control: Git, GitHub, GitLab, DVC
• CI/CD: GitHub Actions, Jenkins, CircleCI

• Data preprocessing & augmentation
• Feature engineering & selection
• Time series analysis
• Signal processing (filtering, FFT, wavelet transforms)
• Data visualization (Matplotlib, Seaborn, Plotly)
• Big data tools (Spark - basic knowledge)

• IDEs: VS Code, PyCharm, Jupyter Lab
• Notebooks: Jupyter, Google Colab, Kaggle Kernels
• Containerization: Docker, Docker Compose
• Orchestration: Kubernetes (K8s), Helm
• Monitoring: Prometheus, Grafana

Master's Thesis | University of Cincinnati

Developed a novel deep learning framework for predicting visual stimuli from brain activity using EEG-fMRI fusion:

Key Contributions:

• Designed and implemented a Temporal Convolutional Network (TCN) architecture for multimodal fusion of brain signals
• Achieved state-of-the-art classification accuracy on visual stimuli prediction from neural data
• Conducted comprehensive survey of 100+ papers on multimodal neuroimaging methods
• Demonstrated effective integration of temporal (EEG) and spatial (fMRI) brain signals with novel fusion strategies

Technical Stack:

• Architecture: TCN with dilated causal convolutions, multi-head attention mechanisms
• Fusion Strategy: Intermediate fusion with learned cross-modal attention weights
• Tools: Python, PyTorch, MNE-Python, Nilearn, scikit-learn, NumPy
• Dataset: Custom EEG-fMRI recordings with visual stimuli paradigm

Impact:

• Novel approach to brain-computer interface applications with potential for real-time decoding
• Applications in neurofeedback, clinical diagnosis, and cognitive neuroscience research
• Demonstrated superior performance over traditional early/late fusion approaches

Thesis Defended: October 2024 | Grade: Excellent
Committee: Prof. Vikram Ravindra (Advisor), Prof. Jun Bai, Prof. Vesna Novak

Production System | Ongoing

Autonomous multi-agent system for end-to-end scientific research automation:

Key Features:

• 🤖 Multi-agent framework using LangChain agents with specialized roles (planner, retriever, synthesizer, critic)
• 💾 Vector database memory system for context retention and knowledge accumulation
• 📚 Automated literature discovery from arXiv, PubMed, Google Scholar with intelligent filtering
• ✍️ Autonomous synthesis of research findings into comprehensive reports with citations

Technical Implementation:

• Agent Framework: Custom LangChain orchestration with ReAct prompting
• LLM Backend: GPT-4 for planning and reasoning, GPT-3.5-turbo for execution tasks
• Tools: Semantic Scholar API, arXiv API, web scraping, PDF parsing
• Memory: Pinecone vector database with semantic chunking and retrieval
• Workflow: Autonomous planning → multi-source retrieval → synthesis → quality validation

Achievements:

• 60% improvement in research efficiency compared to manual literature review
• Processes 100+ papers per research query with intelligent relevance filtering
• Generates publication-ready summaries with proper citations and source verification
• Supports multi-turn conversations for iterative refinement of research questions

Use Cases:

• Rapid literature reviews for new research projects
• Automated survey generation for grant proposals
• Keeping up with fast-moving research fields (GenAI, multimodal learning)
• Educational tool for understanding complex research topics

Healthcare AI System | Production

FHIR-compliant multimodal health assistant for personalized medical recommendations:

Technical Details:

• 🏥 GPT-4 integration with specialized medical knowledge and clinical reasoning
• 📊 FHIR-compliant lab report parsers supporting 15+ standard biomarker categories
• 🧬 Knowledge graph profiling for patient history and medical context
• 🎯 Personalized recommendations based on comprehensive health profile analysis

Architecture:

• Multimodal Input: Text (medical history, symptoms), structured data (lab reports), images (reports, scans)
• FHIR Integration: Standardized health data interoperability for EHR systems
• Knowledge Graph: Neo4j for medical entity relationships and temporal tracking
• LLM Orchestration: GPT-4 with RAG over medical literature and clinical guidelines

Results:

• 20% increase in diagnostic accuracy through comprehensive context analysis
• Automated processing of lab reports across multiple formats (LabCorp, Quest, hospital systems)
• Personalized health insights across cardiovascular, metabolic, liver, kidney, and thyroid markers
• HIPAA-compliant architecture with end-to-end encryption

Tech Stack:

• Python, LangChain, GPT-4, Neo4j, FHIR APIs, PDF parsing, OCR (Tesseract)
• Secure deployment on AWS with data encryption and access controls

Conversational Search System | Production

Real-time AI-powered search engine with source verification and citation:

Approach:

• 🔍 Real-time web crawling with intelligent content extraction and relevance scoring
• 🦙 Llama-based models for efficient on-device inference and privacy
• 📝 Source citation pipeline with automatic fact-checking and verification
• 💬 Conversational interface for iterative query refinement

Technical Stack:

• LLM: Llama 2 13B with 4-bit quantization for efficient deployment
• Web Crawling: Custom Scrapy spiders with JavaScript rendering (Playwright)
• Ranking: Hybrid BM25 + semantic similarity with cross-encoder reranking
• Citation: Automatic source extraction, quotation matching, claim verification

Outcomes:

• 40% reduction in misinformation through rigorous source verification
• Transparent answer generation with inline citations and confidence scores
• Real-time responses with <3 second latency for most queries
• Support for follow-up questions and multi-turn conversations

Differentiation:

• Unlike black-box LLMs, provides full source transparency
• Combines traditional search quality with LLM understanding
• Privacy-focused with option for local-only inference

Deep Learning for Healthcare | Technocolabs Softwares

Advanced computer vision system for medical image analysis:

Approach:

• 🔬 Deep learning algorithms for pattern recognition in medical imaging
• 📊 Ensemble methods combining CNN and transformer architectures
• 🎯 Transfer learning from ImageNet and medical imaging datasets

Tech Stack:

• PyTorch, TensorFlow, ResNet, Vision Transformers (ViT)
• Medical imaging libraries: SimpleITK, PyDicom
• Data augmentation: Albumentations, custom augmentation pipelines

Outcomes:

• 25% improvement in predictive accuracy for clinical decision support
• 45% boost in system efficiency through optimized inference pipelines
• 10% improvement in diagnostic accuracy validated through Power BI dashboards
• Successfully deployed in clinical pilot program

Research & Exploration:

• 🧪 Parameter-Efficient Fine-Tuning Experiments: Comprehensive evaluation of LoRA, QLoRA, and adapter methods on medical NLP tasks
• 📚 Multimodal Fusion Benchmarks: Comparative analysis of early, late, and intermediate fusion strategies across vision-language tasks
• 🎨 Stable Diffusion Fine-tuning: Domain adaptation for medical image generation with LoRA
• 🔊 Audio-Visual Learning: Cross-modal attention for speech recognition and video understanding

Open Source Contributions:

• Contributions to Hugging Face Transformers documentation
• Bug fixes and feature requests for LangChain
• Tutorial notebooks on advanced RAG techniques

"A Multimodal Neuroimaging Method for the Prediction of Visual Stimuli: A Temporal Convolutional Network-Based Approach for EEG-fMRI Fusion"

• University of Cincinnati, College of Engineering and Applied Science | October 2024
• Advisors: Prof. Vikram Ravindra (Chair), Prof. Jun Bai, Prof. Vesna Novak
• Focus: Temporal Convolutional Networks, Multimodal Fusion, EEG-fMRI Integration, Brain-Computer Interfaces
• Key Contributions: Novel TCN-based fusion architecture, comprehensive survey of multimodal neuroimaging methods, superior performance over traditional fusion approaches

1. A Review on Secure Data Transmission for Banking Application using Machine Learning

• Authors: Gurram, B., Reddy, M. D., & Thatikonda, M.
• Journal: International Journal of Engineering and Advanced Technology (IJEAT)
• Volume: 10, Issue 5 | Pages: 182-186 | Year: 2021
• Focus: Machine learning applications in financial security, secure data transmission protocols
• DOI/Link: View Publication

2. Analysis of Big Data Challenges and Different Analytical Methods

• Authors: Gurram, B., & Reddy, M. D.
• Journal: International Journal of Engineering Research and Advanced Technology (IJERAT)
• Volume: 7, Issue 3 | Pages: 33-38 | Year: 2021
• Focus: Big data analytics methodologies, scalability challenges, distributed computing
• DOI/Link: View Publication

3. Challenges and Solutions for Improving SSD Performance

• Authors: Gurram, B., et al.
• Journal: International Journal of Research in Engineering and Science (IJRES)
• Volume: 9, Issue 8 | Pages: 37-39 | Year: 2021
• Focus: Storage systems optimization, performance tuning, system architecture
• DOI/Link: View Publication

"Temporal Convolutional Networks for Multimodal Neuroimaging: An EEG-fMRI Fusion Approach"

• Manuscript based on Master's thesis
• Target venue: IEEE Transactions on Neural Systems and Rehabilitation Engineering
• Status: In preparation for submission

"Raptor RAG: Hybrid Knowledge Graph Integration for Enhanced Retrieval-Augmented Generation"

• Novel RAG architecture with evidence database injection
• Target venue: NeurIPS 2025 or EMNLP 2025
• Status: Experimental phase

University of Cincinnati Graduate Research Symposium

• Presentation: "Multimodal Neuroimaging for Visual Stimuli Prediction using Temporal Convolutional Networks"
• Date: May 2024

Coming Soon:

• "Building Production-Ready Agentic AI Systems: A Practical Guide"
• "Optimizing LLMs with Quantization: GPTQ Deep Dive"
• "Multimodal RAG: Integrating Text, Images, and Knowledge Graphs"
• "From Research to Production: Deploying TCN Models at Scale"

I'm currently working on:

1. 🤖 Agentic AI Systems - Building autonomous agents that can reason, plan, and execute complex tasks
2. ✨ Foundation Model Optimization - Parameter-efficient fine-tuning and deployment of large models
3. 🔬 Multimodal AI Research - Advancing cross-modal learning and fusion techniques
4. 🏥 Medical AI Applications - Translating research into clinical tools
5. 📝 Technical Writing - Sharing knowledge through blog posts and tutorials

🏆 Graduate Incentive Award for Academic Excellence
University of Cincinnati, 2022-2024

• $60,000 scholarship awarded for outstanding academic performance
• Recognized for exceptional research potential and academic achievement
• Maintained 3.95/4.0 GPA throughout Master's program

🎓 Summa Cum Laude
Bachelor of Technology, SRM University

• 3.98/4.0 GPA - Top of class performance
• Graduated with highest honors

📚 MicroMasters in Data Science
UC San Diego, 2022

• Successfully completed advanced graduate-level data science program
• Demonstrated mastery in machine learning, probability, and statistical inference

💼 Production AI Systems

• Deployed systems processing 50M+ document embeddings in production
• 95% automation of clinical research workflows through agentic AI
• $24K annual cost savings through LLM optimization
• 60% improvement in research efficiency through autonomous agents

👨‍🏫 Educational Leadership

• Trained 500+ students in AI, ML, and cloud computing
• Achieved 90% assignment completion rate through innovative teaching methods
• 40% increase in student engagement through interactive curriculum design
• Designed and delivered workshops reaching 95% positive feedback

Microsoft Certified: Azure Data Fundamentals (DP-900)
Microsoft | March 2023

• Cloud data services, core data concepts, analytics workloads
• Azure data services including SQL Database, Cosmos DB, Synapse Analytics

DeepLearning.AI TensorFlow Developer Specialization
Coursera | May 2021

• Deep learning with TensorFlow
• Convolutional Neural Networks, RNNs, NLP
• Sequence models and time series

Machine Learning
Stanford University (Andrew Ng), Coursera | March 2021

• Foundational machine learning algorithms and theory
• Supervised and unsupervised learning
• Best practices in ML project development

IBM Data Science Professional Certificate
Coursera | 2021

• Data science methodology and tools
• Python for data science, data analysis, and visualization
• Machine learning with Python

Applied Data Science I: Scientific Computing & Python (with honors)
WorldQuant University | 2021

• Advanced Python programming for scientific computing
• NumPy, Pandas, data manipulation and analysis
• Completed with honors distinction

👥 Team Leadership

• Led 4-member deep learning team at Technocolabs Softwares
• Trained 5 interns in advanced AI techniques at Zasti Inc
• Delivered 3 comprehensive workshops on AI/ML and agent development

🎯 Mentorship & Teaching

• Graduate Teaching Assistant for 4 semesters
• Mentored students on 50+ AI/ML projects
• Created educational materials reaching hundreds of learners

Always staying on the cutting edge:

• 🔭 Advanced multi-agent coordination and swarm intelligence
• 🌱 Latest developments in vision-language models (GPT-4V, Gemini)
• 🧪 Novel architectures for continual learning
• ⚡ Efficient inference techniques for large models
• 🔒 AI safety and alignment research

I'm always excited to discuss AI, collaborate on projects, or explore new opportunities:

• 💼 LinkedIn: linkedin.com/in/bhaskar-gurram
• 📧 Email: [email protected]
• 📱 Phone: +1 (513) 652-2523
• 📍 Location: Ashburn, VA
• 📚 Google Scholar: Publications and research
• 🔬 ResearchGate: Academic profile

I'm interested in:

• 🤝 Research Collaborations - Multimodal AI, Agentic Systems, Medical AI, Neuroimaging
• 💼 Full-time Roles - Senior AI/ML Engineer, Research Scientist, Applied AI roles
• 🎓 PhD Programs - Machine Learning, AI, Computational Neuroscience, Medical AI
• 📝 Peer Reviewing - ML/AI conferences (NeurIPS, ICML, ICLR) and journals (IEEE, Nature)
• 🗣️ Speaking Engagements - Technical talks on agentic AI, RAG systems, multimodal learning
• 👥 Mentorship - Guiding aspiring AI engineers and researchers

Expertise Areas for Collaboration:

• Agentic AI and multi-agent systems
• Multimodal learning and fusion
• Retrieval-Augmented Generation (RAG)
• Medical AI and healthcare applications
• LLM optimization and deployment
• Responsible and explainable AI

When I'm not training models or debugging agents:

• 📖 Reading about cognitive science, neuroscience, and philosophy of mind
• 🎓 Creating educational content to make AI accessible
• 👨‍🏫 Mentoring students and aspiring AI practitioners
• 🌍 Following developments in AI ethics and responsible AI
• 🏃 Staying active and maintaining work-life balance