I build products end to end with TypeScript, React, Next.js, Node.js and FastAPI, and I like putting LLMs to work in places where the code can still check their answers. At CodeChef-VIT I work on student tools that thousands of people use every semester. I also build ML projects on the security side: anomaly detection on network traffic and container syscalls.
- Building LLM agents with LangGraph, plus RAG pipelines on vector databases
- Exploring ML for security: catching unusual network traffic and container behaviour
- I design too: Figma, wireframes, prototypes
- Ask me about agents that plan while code verifies, RAG, or Next.js
- How to reach me: krishitaguptaa@gmail.com
| Project | What it does | Stack |
|---|---|---|
| FFCS-inator · live | CodeChef-VIT's timetable planner, used by 30,000+ students. I built an AI agent that turns "no 8am, Friday free" into a clash-free timetable. The LLM plans and code verifies every result. | Next.js · LangGraph.js · Gemini · Zod |
| Transaction Risk Platform · live | Scores card transactions for risk and explains each score in plain English. Includes an AI analyst that answers questions and cites the transactions it used. | FastAPI · Next.js · PostgreSQL · MongoDB Vector Search · Claude API |
| Papers by CodeChef-VIT · live | Previous-year exam papers for 10,000+ students. I worked on search, filters and the upload pipeline to cloud storage. | Next.js · MongoDB · Google Cloud Storage |
| FlowLens | Real-time network intrusion detection that explains every alert as traffic arrives, shown on a live dashboard. | Python · LightGBM · Isolation Forest · TreeSHAP · Streamlit |
| Syscall Anomaly Detection | Catches suspicious behaviour inside Docker containers by tracing system calls with eBPF and scoring them with an autoencoder. | Python · eBPF · Docker · Autoencoders |
| MediSlot | Clinic appointments and billing, with all business rules enforced inside the database through PL/SQL, triggers and packages. | Oracle · Node.js · Express · React |
| Lab PC Fit Checker | Checks a project's software needs against lab PC specs and explains why each machine passes or fails. | React · TypeScript |
