name : Amaan Shikalgar
role : Backend & Full Stack Developer
location : Pune, Maharashtra, India
education : B.E. Computer Engineering — Dr. D.Y. Patil IEMR (2021–2025)
experience : Software Engineer Intern @ Sayu Softtech (Feb–May 2024)Backend-focused developer with hands-on experience building production-ready REST APIs, authentication systems, and scalable web applications. Skilled across the full stack — from designing PostgreSQL schemas and MongoDB data models to shipping React frontends. Recently expanded into applied AI, building RAG pipelines with FAISS and LLM integration from scratch.
I care about systems that are reliable, well-structured, and actually deployed — not just running locally.
Open To: Backend Engineer · Full Stack Developer · API Developer · Junior Software Engineer
| Domain | Proficiency | Details |
|---|---|---|
| Retrieval-Augmented Generation (RAG) | ⭐⭐⭐⭐ | FAISS vector indexing, semantic chunking, retrieval-grounded prompting |
| Vector Search | ⭐⭐⭐⭐ | FAISS nearest-neighbour retrieval, embedding generation |
| LLM Integration | ⭐⭐⭐⭐ | Groq API, LLaMA 3, prompt engineering for factual grounding |
| Sentence Transformers | ⭐⭐⭐ | Semantic embeddings, similarity search |
| AI System Design | ⭐⭐⭐⭐ | End-to-end pipeline design, hallucination reduction |
DocuMind AI — RAG-based Document Q&A System
Semantic document intelligence system that enables natural language querying across large PDF documents with page-level citations. Built on a full RAG pipeline engineered from scratch.
| Attribute | Details |
|---|---|
| Stack | Python, FAISS, Sentence Transformers, LLaMA 3, Groq API, Streamlit |
| Scale | 50+ page documents, thousands of vector chunks |
| Performance | Millisecond vector retrieval via FAISS nearest-neighbour search |
| Security | Environment-based API key management, no data persistence |
| Impact | Zero hallucination design — system refuses to answer if context is absent |
| Repository | GitHub · Live |
Pipeline: PDF → Text Extraction → Chunking → Sentence Embeddings → FAISS Index → Query Embedding → Top-K Retrieval → LLaMA 3 → Grounded Answer + Page Citations
The retrieval-grounded prompting architecture ensures the LLM operates exclusively on retrieved document context, eliminating hallucination at the source rather than filtering it downstream.
StockFlow — Full Stack Inventory Management System
Production-grade inventory management platform with a FastAPI backend, React frontend, and PostgreSQL persistence. Features JWT authentication, optimized query performance, and full cloud deployment.
| Attribute | Details |
|---|---|
| Stack | FastAPI, PostgreSQL, React, JWT, Render, Neon, Vercel |
| Scale | 10+ REST API endpoints, multi-user role support |
| Performance | Query optimization reduced API latency from 320ms → 230ms |
| Security | JWT-based authentication, environment-based secrets |
| Impact | End-to-end full stack deployment across three cloud platforms |
| Repository | GitHub · Live |
Architected with a clean separation between the API layer and data layer. Database query profiling was used to identify and eliminate N+1 patterns, yielding a measurable 28% reduction in average response latency.
DevScan API — GitHub Profile Intelligence API
REST API backend that fetches, analyzes, and persists public GitHub profile data using the GitHub REST API. Features intelligent UPSERT logic and authenticated rate limit scaling.
| Attribute | Details |
|---|---|
| Stack | Node.js, Express.js, MySQL, Railway |
| Scale | 3 REST endpoints, persistent profile storage |
| Performance | Rate limit scaled from 60 → 5000 req/hr via GitHub token auth |
| Security | Environment-based config, GitHub token management |
| Impact | UPSERT logic prevents duplicates and supports re-analysis workflows |
| Repository | GitHub · Live |
Designed around idempotent data operations — repeated analysis of the same profile updates existing records rather than creating duplicates, making it safe for scheduled or event-driven execution.
CourseHub LMS — Backend API for Course Platform
Secure backend REST API for a course-selling platform with multi-role authentication, MongoDB data modeling, and protected route architecture.
| Attribute | Details |
|---|---|
| Stack | Node.js, Express.js, MongoDB, Mongoose, JWT |
| Scale | 10+ REST endpoints across user, admin, and course routes |
| Performance | Mongoose schema optimization for efficient document queries |
| Security | Role-based JWT with separate token secrets per role |
| Impact | Custom middleware architecture cleanly separates auth from business logic |
| Repository | GitHub |
Implemented dual-secret JWT architecture — user and admin tokens are signed with independent secrets, meaning a compromised user token cannot be escalated to admin access under any circumstance.
Feb 2024 – May 2024 · Ahmednagar, Maharashtra, India
Contributed across the full software development lifecycle at a software product company, working directly with senior engineers on real application codebases.
- Participated in application development, testing, debugging, and technical documentation across the SDLC
- Translated user requirements into technical solutions, collaborating with senior engineers to investigate and resolve software issues hands-on
- Gained practical exposure to professional software engineering workflows, version control practices, and team-based development
| Recognition | Details |
|---|---|
| 🥇 President, Coders Club | Led 50+ member technical community, organized coding events and workshops |
| 🏅 Smart India Hackathon | Selected among Top 15 teams out of 100+ participating teams |
| 🌐 GDG Cloud Pune | Active contributor to Google Developer Group Cloud Pune chapter |
| 🎓 GDG On Campus WOW | Core member of GDG On Campus technical event organization team |
learning:
- Advanced system design patterns
- Distributed systems fundamentals
- AI/ML pipeline engineering
building:
- Job discovery automation with AI scoring
- Open source backend tooling
exploring:
- Vector databases beyond FAISS
- LLM fine-tuning vs RAG tradeoffs
- Event-driven architecture with message queues
open_to:
- Backend Engineer roles
- Full Stack Developer positions
- Startup engineering teams
- Remote opportunities
"Good systems are built twice — once in planning, once in code."