Open to opportunities

Kesavardhan Makireddi

@kesavardhanmakireddi

Message

AI engineer building LangGraph and RAG systems for multi-tenant platforms.

India

Message

What I'm looking for

I’m looking for a role where I can ship real AI systems end-to-end—RAG, agent orchestration, and production APIs—using FastAPI and cloud deployments, while building automation workflows that are reliable and idempotent.

I’ve been architecting a multi-tenant B2B AI platform at Course5i, using LangGraph with Azure AI Search and OpenAI GPT-4/Claude for multi-model agent routing and dynamic workflow generation.

I build production-grade RAG pipelines with tenant-isolated vector indexing, and I’ve wired up AI orchestration end-to-end with LangChain, WebSocket streaming, and an async FastAPI backend with PostgreSQL, SQLAlchemy, RBAC, and Alembic.

Before that, I did data-science and ML research work—from XML-to-JSON LLM pipelines with LoRA/QLoRA and quantization to a Latent Diffusion-GAN image synthesis training pipeline using PyTorch, Hugging Face Diffusers, and CUDA-accelerated training.

Experience

Work history, roles, and key accomplishments

AI Engineer Intern

Course5i

Oct 2025 - Apr 2026 (6 months)

Architected a multi-tenant B2B AI platform using LangGraph, Azure, OpenAI GPT-4, and Claude with multi-model agent routing and dynamic workflow generation, reducing manual process mapping by 70%. Engineered production-grade RAG pipeline with Azure AI Search and tenant-isolated vector indexing.

LangGraph Azure GPT 4 Claude fastAPI PostgreSQL SQLAlchemy RBAC Alembic React TypeScript Docker CI CD

Data Scientist Intern

Infoteck Solutions

Oct 2024 - Mar 2025 (5 months)

Built an XML-to-JSON transformation pipeline using GPT, Claude, Mistral-7B/70B, and LLaMA, leveraging LoRA/QLoRA fine-tuning and quantization to optimize inference efficiency. Benchmarked foundation model APIs from SambaNova Systems.

GPT Claude Mistral LLaMA LoRAs QLoRA Hugging Face Transformers PEFT Accelerate Workato

Machine Learning Research Intern

Indian Institute of Information Technology, Nagpur

May 2024 - Jul 2024 (2 months)

Architected a Latent Diffusion-GAN framework using PyTorch and LDM, with a VAE-based encoder and CNN discriminator for high-fidelity image synthesis. Built an end-to-end training pipeline using Hugging Face Diffusers and CUDA-accelerated computing.

PyTorch CNN CUDA