Skip to content
View tharunsridhar's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report tharunsridhar

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
tharunsridhar/README.md

Hi, I'm Tharun Sridhar

Full Stack Dev | AI Engineer

Python FastAPI Django JavaScript TypeScript React Node.js LangChain

Full Stack Developer specializing in AI engineering. I build robust Python backends and scalable JavaScript applications, from REST APIs and database integration to RAG pipelines and LLM-powered automation.

  • Full stack: JavaScript / TypeScript / React / Node.js on the web side, Python (Django, FastAPI) on the backend
  • REST API design, database integration, concurrency-safe backends, and role-based access control
  • RAG pipelines and retrieval strategies: chunking, hybrid search, reranking, and failure diagnosis
  • Agentic AI / MCP / LangGraph, from protocol-level tool-calling to human-in-the-loop workflows
  • Test-driven: every flagship project below ships with a real test suite, not a demo script

About Me

B.Tech, Computer Science · VIT, Vellore · 2022–2026

  • Bengaluru, India
  • I like building things that actually run: APIs, agentic pipelines, and concurrency-safe backends, not just notebooks
  • Comfortable going from model training and evaluation through to deployment and wiring the result into a real backend
  • Curious about secure system design: how systems fail in production, and how authorization and architecture choices prevent that

Worked On

  • REST APIs with FastAPI/Django, JWT auth, and relational data modeling
  • Concurrency-safe backends: row-level locking, cache invalidation, idempotent writes
  • Agentic AI systems: MCP servers/clients, LangGraph tool-calling agents, human-in-the-loop workflows
  • Backend systems with clean structure, auditable data, and role-based access

Flagship Projects

Inventra · Role-Based Inventory Management System

Django DRF PostgreSQL Redis Celery JWT

  • Role-based inventory system (Admin / Manager / Employee) with JWT auth and an append-only stock ledger for a fully auditable history
  • Row-level locking fixed a race condition found in Locust load tests: 40 concurrent users lost 16 units of stock before, zero after
  • Redis-cached reports cut 9 queries to 1; Celery handles invoices, low-stock alerts, and nightly reconciliation; deployed on AWS

GitHub · Live Demo


Mr Fix My Home · Local Services Booking Platform

Node.js TypeScript Express PostgreSQL Prisma React

  • Local services booking platform with Customer / Provider / Admin roles and a booking state machine
  • Double-booking prevented at three layers, down to a PostgreSQL exclusion constraint; race tests fire 8 simultaneous requests and exactly one succeeds
  • 62 automated tests, OpenAPI docs, and CI; live on Render + Neon

GitHub · Live Demo


PhotoShare API · Photo & Video Sharing Backend

FastAPI Async SQLAlchemy JWT ImageKit

  • Async FastAPI backend for photo and video sharing, with JWT auth covering register, login, email verification, and password reset
  • Media goes to the ImageKit CDN; only the owner of a post can delete it
  • API and frontend served from a single process, so no CORS setup is needed

GitHub


Finvoro · Personal Finance Management API

Django DRF PostgreSQL JWT pytest

  • Django REST Framework API for multi-account tracking, categorized transactions, monthly budgets, and spending reports
  • JWT auth with per-user data isolation on every endpoint; balances and budget figures computed live, never stored stale
  • Versioned API with Swagger/ReDoc docs and a pytest suite

GitHub


MCP Agent Toolkit · Model Context Protocol Server & Agent, Built from Scratch

MCP FastMCP Groq FastAPI Pytest

  • Real FastMCP server (Tools, Resources, Prompt) running as its own process; the agent reaches it only through an MCP client session over stdio
  • UI shows live discovery results and a per-turn call trace of actual protocol traffic, not a simulation
  • Swapping in any third-party MCP server needs no change to the agent loop; tests drive the real server with no API key

GitHub


Task Manager Agent · LangGraph Tool-Calling Agent with Human-in-the-Loop Safety

LangGraph FastAPI SQLite Groq Pytest

  • LangGraph agent that manages tasks and pauses for human confirmation before any delete (interrupt / resume)
  • Conversation state persisted in a SQLite checkpointer, so context survives a page refresh
  • Tests run the real graph, tools, and interrupt flow with a scripted fake model, no API key; UI includes a raw request/response inspector

GitHub


More Projects

  • RAG Explorer – A RAG debugger: chunking lab, 2D embedding map, side-by-side config comparison, retrieval-vs-generation failure diagnosis, and dense / MMR / BM25 / hybrid strategies. FastAPI Chroma sentence-transformers Groq
  • Restaurant Recommendation System – Multimodal (text + CLIP image) vector search with a multi-agent LangGraph workflow for trend, cuisine, and nutrition-aware recommendations. FastAPI LangGraph ChromaDB CLIP
  • Retriever Showdown – The same question through five retrieval strategies (baseline, MMR, multi-query, BM25, reciprocal rank fusion) side by side. FastAPI LangChain
  • Vector Retrieval Lab – ChromaDB fundamentals: HNSW index tuning, metadata and full-text filtered search, and RAG retrieval. FastAPI ChromaDB Groq
  • Agent Building Blocks – Seven panels covering tool creation, manual tool calling, LCEL, agent memory, a SQL agent, and a data-visualization agent. FastAPI LangChain LangGraph
  • Agent Framework Showdown – One researcher-and-writer task built in CrewAI, AutoGen, and BeeAI. CrewAI AutoGen BeeAI

Open Source Contributions


Skills

Full Stack · JavaScript / TypeScript (React, Node.js, Express) on the web side, Python (Django, FastAPI) on the backend

Languages · Python, TypeScript, JavaScript, Java, SQL

CS Fundamentals · Data Structures & Algorithms, OOP, DBMS, Operating Systems, Computer Networks

Backend · Django, Django REST Framework, FastAPI, Node.js, Express, REST API Design, JWT Auth & RBAC, SQLAlchemy, Alembic, Celery, django-filter

Frontend · React, HTML, CSS

AI Engineering · LLM Integration (Gemini, Groq, OpenAI), Prompt Engineering, RAG (Hybrid Search, Reranking, ChromaDB), RAGAS, Agentic Tool-Calling, MCP, LangChain, LangGraph, Multi-Agent Frameworks (CrewAI, AutoGen, BeeAI)

Databases · PostgreSQL, MySQL, SQLite, Redis, FAISS

DevOps & Cloud · Docker, GitHub Actions (CI/CD), AWS (EC2, ALB, VPC, IAM), Railway, Git, Linux

Testing · pytest, pytest-django, Locust

Also on the live portfolio site.


Certifications

  • AWS Solutions Architect Associate (SAA-C03) Specialization – Packt (Jul 2026)
  • IBM RAG and Agentic AI Professional Certificate · IBM (Aug–Sep 2026)
  • Artificial Intelligence (Credit Course) · SmartBridge × Google for Developers (Jul 2025)

Contact

🔗 LinkedIn 🔗 HackerRank 🔗 LeetCode 🔗 Hugging Face

Pinned Loading

  1. Inventra Inventra Public

    Role-based inventory management backend (Django, DRF, PostgreSQL, Redis, Celery) with row-level-locking-verified concurrency safety and a live AWS deployment.

    Python 1

  2. finvoro-finance finvoro-finance Public

    Personal finance management API and web app built with Django, DRF, and PostgreSQL JWT auth, budgets, transaction filtering, and spending reports.

    Python

  3. mcp-agent-toolkit mcp-agent-toolkit Public

    A full-stack MCP demo: a real MCP server exposing notes/tasks as Tools, Resources, and a Prompt, paired with a Groq tool-calling agent that talks to it only through an MCP ClientSession over stdio …

    Python

  4. restaurant-recommendation-system restaurant-recommendation-system Public

    End-to-end multimodal restaurant recommendation system: unstructured text and food images become a structured, retrievable knowledge base, served by a multi-agent recommendation workflow, exposed t…

    Python