Hi, I'm Mahesh
AI Engineer & Entrepreneur of my own life. I love building things and helping people. Very active on Twitter.
MB

About

I'm a final-year Computer Science student focused on AI engineering, building RAG systems, AI agents, and full-stack products. I spend my time experimenting with new AI technologies, building products from scratch, contributing to hackathons, Open source and turning ideas into working software.

Work Experience

Skills

Python
TypeScript
JavaScriptJavaScript
Java
CC
SQLSQL
DSADSA
React
Next.js
Node.js
FastAPIFastAPI
PyTorchPyTorch
Machine LearningMachine Learning
NLPNLP
Computer VisionComputer Vision
LLMsLLMs
RAGRAG
AI AgentAI Agents
Agentic AIAgentic AI
Agentic WorkflowsAgentic Workflows
Loop EngineeringLoop Engineering
LangChainLangChain
LangGraphLangGraph
MCPMCP
Prompt EngineeringPrompt Engineering
Context EngineeringContext Engineering
QdrantQdrant
REST APIREST APIs
PostgreSQL
SupabaseSupabase
AWSAWS
VercelVercel
GitGit
GitHubGitHub

Tools

Claude Code
CodexCodex
AntigravityAntigravity
OpenCodeOpenCode
T3 CodeT3 Code
CmuxCmux
Paper DesignPaper Design

GitHub Activity

1,488 contributions in the last year

1 year total
1,488contributions
Oct 26, 2025 — Oct 26, 2026
Busiest day
200contributions
Sep 15
Longest streak
60days
Aug 28 — Oct 26
Current streak
60days
Aug 28 — Oct 26
Drag to orbit · double-click to reset
LessMore

My Projects

Check out my latest work

I've worked on a variety of projects, from simple websites to complex web applications. Here are a few of my favorites.

RAG Pipeline with Hybrid Search

Answers questions over internal docs by mixing dense and keyword search, reranking the results, and citing its sources. It refuses to answer when its own confidence score falls below a threshold. Reached 0.923 composite confidence.

Python
Qdrant
BM25
OpenAI API
FastAPI

RAG Trace Debugger

Shows where a RAG pipeline went wrong. A failure localization engine pins the fault to one of five stages and hit 100% accuracy on a 15-case test suite. The tracing library adds 0.42ms average overhead, and a dashboard shows each trace stage by stage.

Python
FastAPI
React
TypeScript
Gemini API

TruthLens

Deepfake detector that went from a 54.1% F1 baseline to over 95% F1, with false positives cut from 49.8% to under 4.8%. Built on a cleaned 190,335-image dataset and runs in under 2 seconds per image. Placed 5th of 120 teams at HACKATTACK 2026.

React
TypeScript
Node.js
Python
PyTorch
Flask
Gemini AI

AI Agent for Farming

Farming advisor in 23 languages (English plus 22 Indian regional languages). Covers 20 crop profiles and pest and disease diagnosis for 28+ crops, with leaf-photo identification and a 0–100% confidence score.

React
TypeScript
Python
Gemini 2.5 Flash
Hackathons

I like building things

During my time in university, I attended 6+ hackathons. People from around the country would come together and build incredible things in 2-3 days. It was eye-opening to see the endless possibilities brought to life by a group of motivated and passionate individuals.

HACKATTACK 2026

HACKATTACK 2026

Hyderabad, India

Developed an AI-powered deepfake detection and media authenticity platform using ResNet18, PyTorch, and image forensics. Built a hybrid analysis pipeline combining ML classification, metadata analysis, Grad-CAM explainability, and confidence calibration to identify AI-generated or manipulated media and generate forensic reports.

HYPERFUSION

HYPERFUSION

JIS College of Engineering, Kalyani, West Bengal, India

Developed an AI-powered Smart Education solution during an 8-hour offline hackathon, focusing on personalized learning, intelligent content assistance, and improved student engagement. Built a functional prototype under time constraints, applying AI and full-stack development to address real-world challenges in education.

HACKVERSE: INTO THE WEB(Online)

HACKVERSE: INTO THE WEB(Online)

TechnoVIT 2026, VIT Chennai, India

Developed and deployed an AI-powered real-world application during the Multiverse Sprint, focusing on product engineering, AI integration, scalability, and user experience. Advanced from rapid prototyping to a functional, deployment-ready product within the 24-hour hackathon.

METAMORPH 2.0(Online)

METAMORPH 2.0(Online)

Guru Nanak Institute of Technology, Kolkata, India

Developed SentinelAI, an AI-powered cybersecurity system for detecting phishing, credential harvesting, domain spoofing, and social-engineering attacks. Built a real-time threat analysis pipeline combining browser signals, URL analysis, DOM inspection, JavaScript analysis, and LLM-based security reasoning.

DSU DEVHACK 3.0

DSU DEVHACK 3.0

Dayananda Sagar University, Bangalore, India

Developed an AI-powered application during a 36-hour national hackathon, applying generative AI and full-stack engineering to solve a real-world problem. Built and iterated on a production-ready prototype under time constraints, focusing on AI integration, system architecture, and user experience.

The Great Agent Hackathon 2026

The Great Agent Hackathon 2026

Bangalore, India

Developed ResolveX, an autonomous customer resolution engine during a 24-hour in-person AI hackathon. Designed a multi-agent system that understands customer requests, routes tasks to specialist agents, retrieves policy context through RAG, executes approved actions using Freshworks Agent Studio and MCP, verifies outcomes, and escalates complex or high-risk cases to human support. The project combined agentic AI, enterprise automation, voice AI through ElevenLabs, and full-stack product engineering into a working prototype focused on resolving customer issues rather than simply answering them.

Contact

Get in Touch

Want to chat? Just shoot me a dm with a direct question on twitter and I'll respond whenever I can. I will ignore all soliciting.

GitHub
LinkedIn
X