Full-stack and backend engineer building production systems — real-time marketplaces, role-based platforms, and AI-native applications powered by RAG and LLM pipelines.
Computer Science undergraduate at IIIT Kottayam who ships complete systems end to end — frontend, backend, database and deployment. Most of my work has gone into production platforms used by real customers: a hyperlocal commerce app with live order tracking, a client-facing laundry booking service processing real payments, and an AI campus assistant built on a retrieval-first RAG architecture. Comfortable moving between React/React Native interfaces, Node/Flask/FastAPI services, SQL and NoSQL data layers, and Gemini-powered AI workflows.
|
Full-Stack Systems End-to-end products — responsive interfaces wired to REST APIs, authentication and databases, shipped to real users rather than left as demos. |
Backend Engineering API design, role-based access control, transaction/payment validation and workflow automation across Node.js, Flask and FastAPI. |
|
Real-Time & Data Systems Firestore/Supabase-backed platforms with live order tracking, security rules, and multi-role data flows that update instantly across clients. |
Applied AI Retrieval-grounded LLM features — RAG pipelines over pgvector, Gemini API integrations, and scikit-learn inference services. |
Languages
C++ · Python · TypeScript · JavaScript · SQL
Frontend
Backend
Databases & Cloud
Tools & DevOps
Applied AI
Gemini API · Retrieval-Augmented Generation · Vector Search (pgvector) · LLM Workflows · Scikit-learn
RAG-grounded campus assistant for IIIT Kottayam — answers are retrieved from Postgres and vector search first, then phrased by Gemini, never invented from model memory.
- Retrieval-first architecture: SQL + pgvector search runs before any LLM call, so Gemini explains grounded data instead of hallucinating
- FastAPI backend with Google OAuth, httpOnly session cookies, and role-based access for students, faculty, CRs and admins
- OCR ingestion pipeline (Tesseract) that turns PDFs, timetables and notices into searchable Gemini embeddings
A four-role marketplace — customer, retailer, delivery partner, admin — with live order tracking over Firestore.
- Firestore
onSnapshotlisteners drive every screen in real time, from inventory toggles to delivery status, with no manual refresh - Role-based Firestore security rules enforce order-state transitions (placed → accepted → packed → picked up → delivered) server-side
- Multi-store cart splits a single order across retailers and resolves delivery eligibility only once every store is ready
Production booking and billing platform running real customer orders for a laundry service business.
- PhonePe payment integration with callback handling and order-status reconciliation
- Firebase-backed order, billing and service-history workflows used in daily operations
- Cloudinary-based invoice generation shared directly with customers over WhatsApp
AI second brain that enriches saved notes with Gemini and answers questions only from what's actually been stored.
- Ingestion pipeline auto-generates summaries and tags on every note via Gemini, backed by a normalized Postgres schema
- Grounded query endpoint retrieves candidate notes by tag overlap and full-text search before asking Gemini to answer
- Public query API route for programmatic, infrastructure-style access to the knowledge base
ML and AI platform that predicts salary ranges, scores resumes, and plans career switches.
- Trained scikit-learn regression model served through a Flask inference API
- Resume parsing (PyMuPDF/mammoth) feeding a skill-gap analysis and AI-generated resume score via Gemini
- Career-switch planner mapping current role and skills to salary deltas and required skills
Freelance Software Developer — VK Wash · Smart Tech Solutions · Bhoomi
2024 – Present
Shipped production web and mobile applications for business clients — backend APIs, authentication, database design and PhonePe payment integration.
Software Project Contributor — IBM SkillsBuild, Edunet Foundation
Jun 2025 – Jul 2025
Contributed to SalaryPilot — ML feature development, debugging, and AI/LLM integration as part of an industry-aligned software project.
Technical Volunteer — Google Developer Student Club
Jan 2025 – Present
Cross-functional technical support for community-run events and initiatives.
TechSprint Winner — Google Developer Student Club
Won for pitching the campus platform that became ORION — unifying events, academics and AI-assisted information access for IIIT Kottayam.
|
|
|
Indian Institute of Information Technology Kottayam
B.Tech, Computer Science and Engineering · CGPA 8.32/10 · Aug 2024 – Present
Open to software engineering roles and ambitious collaborations.