OPEN TO 2027 SDE / RESEARCH ROLES

I'm Shivrajsinh Bhosale

Final-year AI + Chemical Engineering dual major at IIT Gandhinagar, building production-grade software — RAG platforms, semantic search, real-time collaboration tools, and test automation. Previously ASCENT intern at Axis Bank (Automation for TCoE).

view projects download resume
Python · Java · C++ FastAPI React · Node.js PostgreSQL RAG · FAISS Docker
agent.status — bash
// agent.status
role = "Software Engineer · AI"
edu = "IIT Gandhinagar · 2022–2027"
stack = ["Python", "Java", "FastAPI"]
repos = 24
stars = 6
status = "open_to_work"
// 01 — ABOUT

About Me

The short version: I like building software that holds up in production.

Shivrajsinh Bhosale

I'm a senior undergraduate pursuing a dual major in Artificial Intelligence and Chemical Engineering at IIT Gandhinagar, focused on software development. I enjoy designing backend systems, retrieval pipelines, and developer tooling that are fast, reliable, and measurable.

As an ASCENT intern at Axis Bank, I built a Java + Selenium test recorder with a self-healing replay pipeline that cut manual test authoring from hours to under two minutes. My recent work includes a multi-tenant RAG platform with FastAPI and PostgreSQL, a semantic product search engine, and a real-time collaborative code editor — working across Python, Java, C++, JavaScript, and SQL.

  • NAMEShivrajsinh Bhosale
  • LOCATIONGandhinagar, India
  • EDUCATIONIIT Gandhinagar
  • FOCUSSoftware Engineering · AI
  • EMAILshivrajsinh.bhosale@iitgn.ac.in
  • STATUSOpen to opportunities
// 02 — EDUCATION

Academic Background

2022 — 2027

B.Tech, Dual Major
AI and Chemical Engineering

IIT Gandhinagar

CPI: 8.00 / 10

2021 — 2022

Class XII — PCM

Dr. V.G.I. Paranjape, Rahimatpur

88.33%

2019 — 2020

Class X

Podar International School, Satara

95.20%

// 03 — PROJECTS

Featured Work

Selected projects from coursework, competitions, and research. Live on GitHub.

// PROJECT_01
Jul–Aug '26 Python

Multi-Tenant Research Automation Platform

AI research platform for document ingestion, evidence retrieval, grounded answers, and web fallback. Multi-tenant RAG on FastAPI with PostgreSQL RLS and Google OAuth 2.0; HNSW + BM25 hybrid retrieval with RRF lifted Recall@5 from 74% to 91% on MS MARCO / SciFact. CRAG with Groq / Tavily behind Nginx load balancing cut P95 latency 30%+ at 450+ req/s.

FastAPI PostgreSQL RAG Nginx
repo live
// PROJECT_02
Mar–Apr '26 Python

E-Commerce Semantic Search Engine

Semantic search over 12,732 Flipkart products using all-MiniLM-L12-v2 embeddings in FAISS (with Prof. Anirban Dasgupta). Four retrieval models spanning graph traversal, metadata re-ranking, and LLM query expansion, benchmarked via a Flask API — 0.897 NDCG on semantic queries.

FAISS Embeddings Flask Information Retrieval
repo
// PROJECT_03
May–Jun '24 JavaScript

Real-Time Code Collaboration Platform

Collaborative online code editor with real-time multi-user editing over WebSockets — room-based sessions, live code sync, syntax highlighting, and session state management. Frontend on Vercel, backend on Render.

React.js Node.js · Express WebSockets
repo live
// PROJECT_04
Dec '24 Jupyter

FedEx — Optimal Cargo Management for Flights

Inter-IIT Tech Meet 13.0 (IIT Bombay). Sequential Genetic Algorithm with a Deepest-Bottom-Left heuristic for ULD packing, using Modified OX, 2-OPT / orientation mutation, and constraint validation. Converged in ~70 generations to a total cost of 32,936 — within 1,336 of optimal — at 78.86% mean volumetric efficiency.

Genetic Algorithms Bin Packing Inter-IIT
repo
More repositories on GitHub
→ view all on GitHub
// 04 — BLOG

Writing

Paper notes, project write-ups, and the occasional rant.

QKV
JUN 25, 2026 · 12 MIN · #transformers

Attention Is All You Need — Notes from a Slow Read

What the 2017 paper actually says, why it worked, and what surprised me on a careful re-read in 2026.

read post →
soon
COMING SOON · #rag

Hybrid Retrieval in Practice: HNSW + BM25 with RRF

How fusing dense and sparse retrieval took Recall@5 from 74% to 91% — and what it cost in latency.

in draft →
soon
COMING SOON · #optimization

Why Most "AI for Logistics" Demos Don't Survive Contact With a Warehouse

Notes from the FedEx cargo-packing project — modelling tricks that worked and what I'd do differently.

in draft →
0AWARDS
0PROJECTS
0LEADERSHIP ROLES
0INTERNSHIPS
// 05 — RESUME

Resume

Embedded below or downloadable as PDF.

download resume.pdf
// 06 — CONTACT

Get In Touch

Open to software engineering and AI / ML engineering roles. Always happy to chat.

Email

shivrajsinh.bhosale@iitgn.ac.in

Phone

+91 8600 563 130

LinkedIn

connect

GitHub

@xsinh19