Agentic AI01
TripMind
Multi-agent travel planning system using LangGraph, LangChain, Gemini, Tavily search, flight tools, and PostgreSQL-backed state persistence.
LangGraphPythonGeminiPostgreSQL
View repository ↗
Python Developer · Agentic AI Engineer
Building useful software with Python and AI.
Backend systems, AI agents, and full-stack products.
01 / Selected work
Projects that show how I build.
Python systems, agentic AI, search, ML, and full-stack products.
Search / IR02
Scout
Local documentation search engine with Python, Flask, SQLite, inverted indexing, ranking, snippets, highlighting, and section-aware retrieval.
PythonFlaskSQLiteBM25-style ranking
View repository ↗
Product03
Attendance Manager
Attendance management product being built incrementally with a modern Next.js UI, daily attendance flows, and persistent local state.
Next.jsTypeScriptReactProduct UI
View repository ↗
Full Stack04
QuickShow
Movie booking platform with Next.js, TypeScript, Supabase, authentication, real-time seat availability, PostgreSQL data models, and admin workflows.
Next.jsSupabasePostgreSQLAuth
View repository ↗
Security05
PassNest
Secure password manager and API using FastAPI and Svelte, with JWT authentication, cryptographic protection, and password-strength analysis.
FastAPIPythonSvelteJWT
View repository ↗
Machine Learning06
Transaction Fraud Detection
Fraud-scoring service with Flask, XGBoost, and scikit-learn for validated real-time risk classification and batch inference workflows.
PythonXGBoostFlaskPandas
View repository ↗
02 / Agentic AI workflow
From prompt to action, not just prompt to text.
The engineering pattern I use for agentic systems.
Understand
Parse the user's goal, constraints, context, and required output.
Plan
Break the request into a stateful sequence of tasks and decisions.
Retrieve
Bring in trusted context using RAG, search, databases, or APIs.
Use tools
Call external tools such as search, data, or application services.
Act
Generate the next action, structured result, or product-side change.
Validate
Check tool outputs, schemas, business rules, and failure paths.
Remember
Persist useful state so multi-turn workflows can continue reliably.
03 / Technical stack
What I work with.
Python & Backend
Python · FastAPI · Flask · Asyncio · REST APIs · WebSockets · Pydantic · SQLAlchemy · Celery · Redis · PyTest
Agentic AI & RAG
LangGraph · LangChain · Multi-agent systems · Tool calling · RAG · Hybrid search · Vector databases · Gemini · OpenAI · Claude · Guardrails
Data & ML
PostgreSQL · SQLite · Supabase · Pandas · NumPy · scikit-learn · XGBoost · ETL · Data pipelines
Frontend & Product
TypeScript · JavaScript · React · Next.js · Svelte / SvelteKit · Tailwind CSS · Responsive UI
DevOps & Tooling
Docker · Git · GitHub · Linux · Bash · CI/CD · Postman · Vercel · Cloud deployment
AI Development Tools
Codex · Cursor · Claude · OpenCode · Antigravity
04 / Contact
Have a Python, backend, or agentic AI problem to solve?
I'm open to full-time roles, internships, and serious engineering collaborations.