Kranthi Gadi | AI Engineer & Agentic Systems

01

Who I Am

I’m Kranthi Gadi, an AI Engineer in the making with a 4.0 GPA in Applied AI from Purdue University.

I don't just train models; I build systems. My work focuses on solving the "amnesia" problem in AI agents and deploying scalable, containerized ML applications. I thrive at the intersection of deep learning research and production DevOps.

Education & Coursework:

M.S. Applied Artificial Intelligence (Expected Dec 2026)

AI Agents, Machine Learning, Deep Learning
Ethical AI, Generative AI, Data Visualisation
High Performance Computing & Big Data
Special Problems in AI

4.0 CGPA

4+ Agent Systems

AWS Certified Cloud

02

Tech Stack

Large Language Models:

Model Context Protocol (MCP) LangChain LangGraph CrewAI LlamaIndex Hugging Face Transformers

Deep Learning:

PyTorch TensorFlow Keras FastAI

Vector Databases:

Pinecone pgvector ChromaDB Weaviate

Computer Vision:

OpenCV YOLOv8 Detectron2 Stable Diffusion

NLP:

SpaCy NLTK BERT/RoBERTa OpenAI API

Core Languages:

Python (Expert) C++ SQL Bash / Shell

Web Ecosystem:

JavaScript (ES6+) TypeScript HTML5 / CSS3

AWS Cloud:

EC2 Lambda S3 SageMaker IAM

Google Cloud:

Vertex AI Cloud Vision BigQuery

Infrastructure:

Docker Kubernetes GitHub Actions (CI/CD) Linux System Admin

Backend:

FastAPI Flask Django

Data Processing:

Pandas NumPy Apache Spark Hadoop

Visualization:

Matplotlib Seaborn Power BI Tableau Plotly

Databases:

PostgreSQL MongoDB MySQL Redis

Development Tools:

VS Code JupyterLab Git / GitHub Postman

Methodologies:

Agile / Scrum System Design Design Patterns

03

Projects

Currently Architecting

SourceMind

Collaborative AI Memory Infrastructure with Attribution

Overview: Shared persistent memory engine for AI teams, solving the "trust gap" by tracking human vs. AI contribution.
Engineering: Built on Postgres + pgvector with a custom attribution algorithm using edit-distance metrics. RBAC included.
Impact: Turns ephemeral chat context into a queryable knowledge graph with full decision lineage.

FastAPIpgvectorLangChainOpenAIAttribution Algo

Currently Architecting

AgentOS

MCP-Native Operating System for Autonomous Agents

Overview: A research-grade OS treating AI agents as processes and tools as system resources, enabling secure, long-running workflows.
Engineering: Implemented an OS Kernel for process scheduling and a Hierarchical Memory (L1-L4) system using MCP.
Impact: Benchmarked against LangChain/CrewAI, demonstrating superior state consistency and safety in multi-agent orchestration.

PythonModel Context Protocol (MCP)Hierarchical MemorySystem Design

Research Paper

Agentic Purchase System

Vision-Driven Autonomous Commerce with Trust Verification

Overview: A 5-stage orchestration pipeline (Vision → Intent → Checkout) that interprets diverse user inputs to execute autonomous, verified purchases.
Engineering: Powered by LangGraph & FastAPI using a Saga pattern for state management. Features a specialized Trust Agent utilizing Z-score heuristics and LLMs to detect replicas.
Impact: Validated on the Amazon Berkeley Objects dataset, achieving 0.93 Intent F1 and high precision in detecting counterfeit listings.

LangGraphFastAPIGoogle Cloud VisionLLM Orchestration

Experimental & Open Source

Connecting to GitHub...

04

Experience

Graduate Research Assistant

Purdue University

March 2025 - Present

Agentic Systems: Developed an "Agentic Purchase System" to automate complex procurement workflows using multi-agent architectures.
AI Memory Research: Conducting active research under three professors, focusing on long-term memory and state persistence for LLMs (2 concurrent projects).

DevOps Engineer

SynchroServe Global Solutions

Oct 2024 - Dec 2024

AWS Infrastructure: Managed scalable cloud environments using EC2, S3, VPC, and Auto Scaling groups for high availability.
CI/CD Pipelines: Orchestrated automated deployments with Jenkins, Kubernetes, and Git/GitHub, optimizing the SDLC for rapid delivery.
Storage & Compute: Configured EBS/EFS storage solutions and load balancing to ensure robust system performance.

Part-time Instructor (DSA)

Learn Everything AI

March 2024 - June 2024

Technical Education: Created comprehensive video curriculum for Data Structures & Algorithms, simplifying complex logic for students.
Content Strategy: Produced high-quality technical tutorials covering both fundamental and advanced CS concepts with clear, engaging delivery.

05