About Me
I build AI systems that make real-world services accessible to everyone.
I'm Praneeth, an AI & Data Science student at Karunya Institute of Technology and Science, driven by one goal: building systems that work in the real world. My work spans computer vision, NLP, and end-to-end ML pipelines - from raw data to deployed product.
I've shipped a tutoring platform to a paying client, built a gesture-controlled presentation system using OpenCV and MediaPipe, and developed an AI coding assistant using RAG and LangChain. I also built a fisherman border-detection system as a proof of how AI can protect lives.
Status
B.TECH 2nd year at Karunya Institute of Technology and Sciences
Core Stack
Python, SQL, Machine Learning , MERN
Focus
Deployed AI systems
Technical Proficiency
AI & Machine Learning
Frameworks & Libraries
Languages
Web & APIs
Domains & Concepts
Data & Backend
MLOps & System Design
Featured Projects
TruBill — WhatsApp-Native GST Invoicing SaaS
Multi-tenant GST invoicing platform built for small textile and footwear retail shops, with invoices delivered straight to a customer's WhatsApp — zero app download required.
- GST-compliant invoicing (GSTR-1/3B, ITC, credit notes) · barcode/Code128 label generation
- WhatsApp Cloud API delivery · Supabase RLS for multi-tenant isolation · live paying shops in Tiruppur/Coimbatore
LexiCore — Hybrid RAG Platform
An enterprise-grade retrieval-augmented generation system combining dense and sparse retrieval for higher-accuracy document Q&A than vector search alone.
- Hybrid retrieval: BM25 + dense embeddings fused via Reciprocal Rank Fusion (RRF)
- RAGAS-based evaluation pipeline · Dockerized with full CI/CD via GitHub Actions
AgentOps — Multi-Agent Task Orchestrator
A production-grade orchestration system that coordinates multiple AI agents through stateful, resumable task graphs.
- LangGraph StateGraph for agent coordination · async Postgres checkpointing for crash-safe resumability
- Docker SDK integration for sandboxed execution · Streamlit ops dashboard · CI/CD via GitHub Actions
MSME Credit Score Engine
An ML-driven credit scoring model built on TruBill's own invoicing schema, designed to give small merchants a path to formal credit access.
- Random Forest classifier at 95.8% accuracy on 6 engineered financial-behavior features
- SHAP explainability layer · Next.js dashboard with CSV upload and recharts visualization
EchoNode — AI API Gateway with Semantic Caching
A gateway layer that sits in front of LLM APIs, cutting cost and latency through semantic response caching and high-availability routing.
- Semantic caching via Redis to deduplicate near-identical prompts
- Streaming response interception · telemetry layer · HA fallback routing · full CI/CD pipeline
Z-Axis — AI-Powered 2D-to-3D Asset Converter
A pipeline that converts flat 2D images into interactive 3D assets, combining background removal with generative 3D reconstruction.
- Background removal via rembg · 3D mesh generation via TripoSR (HuggingFace Gradio)
- Interactive 3D viewer built with React Three Fiber · Vite/TypeScript frontend
Get In Touch
Send a Message
Fill out the form and I'll get back to you
Connect With Me
Find me on social platforms