I describe myself as a Backend and AI/ML-focused Software Engineer with hands-on experience building REST APIs, RAG systems, LLM-powered applications, and ML pipelines. My portfolio presents work across backend engineering and AI/ML, and I am seeking Backend Engineering or AI Engineering roles.
I have solved 150+ LeetCode problems.
My portfolio highlights 45K+ movies in CineMatch.
My portfolio highlights a 101.5h DS, ML, DL & NLP bootcamp.
I built and deployed my projects independently and learned through official documentation. My project portfolio includes RAGBot, CineMatch, and TableFlow; I describe the work as independently built and deployed.
For RAGBot, I architected a multi-tenant B2B SaaS platform for PDF-based chat with an embeddable JS widget using Django, DRF, PostgreSQL, Celery, and Redis.
I engineered RAGBot’s LangChain pipeline using FAISS, cosine similarity, and SentenceTransformers embeddings; it served contextual answers via GPT-4o-mini with source attribution.
I built RAGBot’s per-tenant document ingestion, chunking, embedding, and async processing pipeline and containerized it with Docker.
I deployed RAGBot on AWS EC2 with a custom domain and HTTPS; its live demo was publicly accessible.
CineMatch combines plot overview, genre, and tagline into a unified text corpus for content-based movie recommendations.
My CineMatch NLP pipeline uses lowercasing, regex cleaning, stop-word removal, and lemmatization.
CineMatch uses TF-IDF vectorization with unigrams and bigrams and 50,000 features, generating a sparse matrix with 1.7M+ non-zero entries; it uses cosine similarity for item-to-item recommendations and serializes trained artifacts for reuse.
I built TableFlow as a QR-based restaurant ordering backend with menu browsing and order lifecycle management: pending → preparing → served.
I implemented JWT authentication, role-based access control, and Docker-based deployment for TableFlow’s admin and customer roles.
My listed programming languages are Python, JavaScript (Node.js), TypeScript, SQL, C.
My backend and API skills include Django, DRF, Flask, FastAPI, Express.js, REST APIs, JWT Auth.
My AI and ML skills include RAG, LangChain, Embeddings, Semantic Search, NLP, Scikit-learn, Classification, Regression.
My vector-search skills include FAISS, ChromaDB, Cosine Similarity.
My database skills include PostgreSQL, MySQL, MongoDB.
My listed tools include Git, AWS, Docker, Postman, Jupyter Notebook, Linux, HuggingFace, Celery, Redis.
My CS fundamentals include Data Structures & Algorithms, Operating Systems, DBMS, OOP, Computer Networks, System Design.
I am pursuing a B.Tech in Computer Science and Engineering at Parul University; the portfolio lists 2023 – 2027 (Expected), a CGPA of 7.65/10, and Vadodara, Gujarat.
I completed Higher Secondary (Class XII), CBSE, at YMS Sr Sec Public School; the portfolio lists 2022 – 2023 and Dhanaura, UP.
I completed KRISHAI Technologies’ Data Science, Machine Learning, DL & NLP Bootcamp in Aug 2026; it was a 101.5-hour bootcamp covering Data Science, Machine Learning, Deep Learning, and Natural Language Processing.
I completed IBM SkillsBuild AI Fundamentals in Sep 2025. It covered fundamentals of machine learning, deep learning, NLP, computer vision, and AI workflows using IBM Watson Studio.
My contact email is Shahjanali0022@gmail.com.
My DocuMind repository is a Python project on GitHub. Started 2026-11, last updated 2026-11.
DocuMind’s intended workflow is: Upload PDFs, ask grounded questions with citations, and extract structured fields.
I built DocuMind as a reusable document-intelligence and RAG backend for company knowledge kept in PDFs such as invoices, policies, and SOPs. It addresses the limits of keyword search for paraphrases such as “payment terms” versus “remuneration schedule,” pure vector search for exact identifiers such as “INV-2024-118,” and answers without citations.
DocuMind runs each query against FAISS for semantic search and BM25 for keyword search in parallel. Reciprocal Rank Fusion merges the rankings with no tuning.
Each hit is tagged dense / bm25 / both so you can see why it matched.
Grounded Q&A uses retrieved chunks as its source: the LLM must cite every claim with markers such as [1] and [2], and says it does not know when it cannot answer from the retrieved material.
Structured extraction uses with_structured_output to turn documents into typed JSON; invoice fields are supported out of the box, and callers can pass any Pydantic model.
Each upload gets its own FAISS index + BM25 corpus under storage/<doc_id>/, lazy-loaded. Adding docs never rebuilds others.
Model, chunking, top-k and fusion settings all live in app/config.py.
The backend uses FastAPI in app/, with a Streamlit demo UI in ui/. The implementation uses gemini-3.6-flash through LangChain, local all-MiniLM-L6-v2 embeddings, PyPDFLoader ingestion, and RecursiveCharacterTextSplitter chunking.
API: POST /documents/upload → POST /ask → POST /extract. See /docs for the interactive Swagger UI.
pip install -r requirements.txt
cp .env.example .env # add your GOOGLE_API_KEY
uvicorn app.main:app --reload
# demo UI:
streamlit run ui/streamlit_app.py
Tests run with pytest tests/ -v. Ingestion and hybrid tests run offline; Q&A and extraction tests need GOOGLE_API_KEY and are skipped without it.
My GitHub profile presents my work and interests alongside the projects I build. My repositories span AI and RAG applications, backend APIs, web applications, learning projects, and other experiments.
My GitHub profile tagline is “I architect RAG systems, write backend APIs, and chase the AI/backend overlap.”
My GitHub profile lists my location as Vadodara, Gujrat.
My GitHub profile says I have been on GitHub since 2025 and have 2 followers.
The language totals shown across the repositories are Python (10), JavaScript (5), HTML (4), Jupyter Notebook (2), TypeScript (2), CSS (1).
My GitHub lists VerifyCOD: Python. Last updated 2026-11.
My GitHub lists Zuno-studio: AI-powered website builder that transforms simple prompts into beautiful, responsive websites. Python. Last updated 2026-11.
My GitHub lists TeddyPups-Clone: HTML. Last updated 2026-11.
The README records the project dates as: Started 2026-10, last updated 2026-11.
My assignment description is: A pixel-perfect clone of the TeddyPups website (puppy.instabizweb.com), built as part of a technical assignment for Insta Biz Web.
The README describes the site this way: TeddyPups is a single-page storytelling website for a premium puppy brand.
The project stack is Next.js 15 (App Router, Turbopack), React 19, Tailwind CSS, and Framer Motion.
The interactive features are: Pinned horizontal scroll-driven kennel with swinging doors and walking mascot; canvas-based paw cursor trail; scroll-direction-aware navbar; smooth FAQ accordion with ARIA attributes.
The README describes the experience as including a loader intro, hero section, journey timeline, passport stamps, happy families gallery, and CTA sections.
The project uses pre-generated responsive images, with no runtime optimizer needed.
I used this prompt to debug the animation engine: “The page loads but React never hydrates — here are the console errors and the bootstrap code, what's breaking?”
I used this prompt for the static-hosting image-loader issue: “Next.js Image component generates /_next/image URLs that 404 on static hosting — how do I override the loader to use existing files?”
I worked through understanding the component structure, scroll engine, and animation timing myself by reading the code.
The image hydration mismatch was that the SSR'd HTML had working image URLs, but React re-rendered them as optimizer URLs on hydration.
I built the project as a static site.
The project credits me as Shahjan Ali — B.Tech CSE '27, Parul University.
The project structure includes: index.html, puppies.html, puppies/ individual detail pages, contact.html, faq.html, how-it-works.html, happy-families.html, gone-home.html, and static asset directories for chunks, images, brand, media, UI graphics, and cursor assets.
The deployment model is: No build step needed — this is a static export. Any static host works — Netlify Drop, Vercel, GitHub Pages, etc.
The local serving instructions are: python3 -m http.server 8000; npx serve .; http://localhost:8000; don't open index.html directly via file:// — the JS chunks need HTTP to load properly.
My AI tools and their use were: Claude (Anthropic) and Muse were used during development; Claude helped debug the animation engine and review patches, while patches were sanity-checked for syntax errors before pushing.
I fixed the bootstrap by: Corrected broken chunk paths after the downloader saved error pages as .html, and deferred the push to window load.
I debugged the inline script corruption as follows: A broken escape sequence (\" lost its backslash) killed an inline script; found by running each script through a JS parser one by one.
I fixed static image loading by: Patched the default image loader to map optimizer URLs to pre-generated static files, fixing static-hosting 404s and hydration URL changes.
My GitHub lists DocuMind: Python. Last updated 2026-11.
My GitHub lists Dental-Assisrant: Python. Last updated 2026-11.
I built Dental-Assisrant as a voice-first dental assistant proof of concept for the VI-SCAN Diagnostics AI Engineer take-home. Its details, design decisions, setup, and limitations are documented on dental assisrant.
My GitHub lists langgraph-learning: Jupyter Notebook. Last updated 2026-10.
My GitHub lists langgraph-chatbot: Python. Last updated 2026-10.
My GitHub lists shoes: just testing a shoes website. HTML. Last updated 2026-10.
My GitHub lists my-portfolio: HTML. Last updated 2026-10.
My GitHub lists CineMatch: ML + NLP powered movie recommendation system using content-based filtering, TF-IDF, and cosine similarity to deliver personalized movie recommendations. Jupyter Notebook. Last updated 2026-10.
My GitHub lists RAG-Bot: Python. Last updated 2026-09.
My GitHub lists Support-agent: AI customer support agent for AmazonHelp that classifies customer intents, retrieves historically similar support interactions, drafts grounded responses, and decides when to escalate to a human. Includes reproducible evaluation, golden-set benchmarking, and RAG baselines. Python. Last updated 2026-09.
My GitHub lists Youtube-chatbot: YouTube RAG Chatbot | Ask questions and get summaries from YouTube videos using LangChain and RAG. Python. Last updated 2026-09.
My GitHub lists Mini-ERP-CRM: TypeScript. Last updated 2026-08.
My GitHub lists Shahjan00: Last updated 2026-06.
My Shahjan00 profile says I am building RAG pipelines that ship to production.
My TableFlow stack includes Django, DRF, PostgreSQL, JWT, and Docker.
My Shahjan00 GitHub profile identifies me as an AI Application Engineer based in Vadodara, Gujarat, and says I am a B.Tech Computer Science student in the 7th semester.
My Shahjan00 profile says I am self-taught through official docs and have solved 150+ LeetCode problems.
My profile says I am currently chasing the gap between "backend dev" and "ML engineer" — and building things that live in that gap.
My profile says I am open to AI Engineer / Backend / Founding Engineer roles.
TableFlow is my QR-Based Restaurant Ordering System, dated Aug – Sep 2025; customers scan a table's QR code, browse the menu, and place an order, with a full lifecycle from pending → preparing → served.
TableFlow uses JWT auth with role-based access control for admins and customers.
TableFlow uses normalized PostgreSQL models via Django ORM; custom middleware for consistent structured error responses.
TableFlow is containerized for one-command deployment.
My URL Shortener is in progress; generates and resolves short URLs through a clean web interface. I am building it to explore Django's request/response cycle, routing, and persistence layer outside of the RAG/SaaS stack.
The Shahjan00 repository profile says it started 2026-06 and was last updated 2026-06.
My profile lists Python, JavaScript, and SQL among my languages.
My profile lists Django, DRF, FastAPI, and LangChain among my frameworks and tools.
My profile lists FAISS, ChromaDB, OpenAI API, and HuggingFace among my AI tools.
My profile lists PostgreSQL, MySQL, and MongoDB among my databases.
My profile lists Docker, Celery, Redis, and Git among my tools.
My GitHub lists AI-Application-Roadmap: JavaScript. Last updated 2026-06.
My GitHub lists track: JavaScript. Last updated 2026-06.
My GitHub lists Node_task-management: A secure, data-driven RESTful API for task management with authentication and dual database integration (PostgreSQL for users, MongoDB for tasks). JavaScript. Last updated 2026-04.
My GitHub lists Open-Book: AI-powered full-stack book intelligence platform using a RAG pipeline. Scrapes and stores book data, generates embeddings, and enables context-aware Q&A via LLMs. Built with Django REST, React, MySQL, and vector DBs, showcasing backend, AI integration, and scalable system design. Python. Last updated 2026-04.
My GitHub lists TableFlow: Multi-tenant QR-based restaurant ordering system built with Django, allowing customers to scan table QR codes, browse menus, and place orders seamlessly. JavaScript. Last updated 2026-04.
My GitHub lists minor-clg-project: TypeScript. Last updated 2026-04.
My GitHub lists Open-Bank: A Bank Transaction System built with Node.js, Express, and MongoDB for secure account and transaction management. JavaScript. Last updated 2026-04.
My GitHub lists URL-shortener: Django project for URL shortener website. CSS. Last updated 2026-03.
My GitHub lists TodoApp: A full-stack Todo web application built with Django featuring user authentication, task CRUD operations, filtering, and search. Created as part of my backend development portfolio. Python. Last updated 2026-03.
My GitHub lists Django-Blog: HTML. Last updated 2026-03.
I’m a B.Tech Computer Science Engineering student specializing in Cyber Security at Parul University, graduating in 2027. I’m in my final year, working on AI-powered projects, improving my backend engineering skills, and preparing for internships and full-time software engineering opportunities. My direction is AI Application Engineering or Backend Engineering, especially work that combines practical AI applications with Python backend systems.
I’m focused on becoming an AI Application Engineer or Backend Engineer, with a strong interest in building real-world applications powered by LLMs, AI agents, and backend systems.
I work primarily with Python, Django, FastAPI, LangChain, LangGraph, and PostgreSQL.
I enjoy building practical software that solves real problems rather than creating projects just to add technologies to my resume.
RAGBot is my multi-tenant B2B PDF-based RAG SaaS platform. Businesses upload documents, which are ingested, chunked, and embedded for FAISS semantic search; the application generates context-aware answers with source citations. Business-specific configurations and customizable chatbot tones support separate tenants, and an embeddable JavaScript widget lets businesses add their chatbot to a website. Its stack includes Python, Django, Django REST Framework, PostgreSQL, Celery, Redis, LangChain, FAISS, Sentence Transformers, OpenAI GPT-4o-mini, React, AWS EC2, Nginx, and Gunicorn. I deployed it at https://ragbot.shahjan.online/.
I benchmarked RAGBot with 200 concurrent users and 20 workers, achieving 180 successful requests and identifying timeout and throughput bottlenecks.
I’m building Zuno, an AI-powered website builder that generates websites from natural-language prompts. Its intended workflow analyzes requirements, plans a website, generates architecture and HTML, CSS, and JavaScript, tests the result in a browser with Playwright, evaluates and fixes bugs, and supports natural-language editing. I’m also implementing conversational memory, summarization, human-in-the-loop approvals, and workflow tracing. The stack includes React, Python, FastAPI, LangGraph, Gemini, PostgreSQL, Docker, Playwright, and LangSmith. The repository is https://github.com/Shahjan00/Zuno-studio.git.
Zuno-studio is described as an AI-powered website builder that transforms simple prompts into beautiful, responsive websites.
The Zuno-studio repository is a Python project on my GitHub, started in 2026-10 and last updated in 2026-11.
The README describes Zuno as an AI website builder: I can describe a website in plain words, and Zuno plans it, breaks the work into tasks, and writes code file by file.
For a build, Zuno’s router selects the build path, the planner makes a project plan and design brief, the architect splits work into one task per file, and the coder writes each file.
Generated project files live in backend/generated_project/.
Zuno stores conversations in backend/zuno.db by default; setting DATABASE_URL allows using Postgres instead.
To run the backend from the repository root, create and activate a virtual environment, install backend/requirements.txt, copy .env.example to .env, and put GROQ_API_KEY in the environment file. Then run the backend and start the frontend in a separate terminal. The README’s local frontend URL is http://localhost:5173; its API documentation host is redacted.
The README’s example chat request is POST /api/chat with {"prompt": "Build a portfolio website"}.
- The README lists
GET /api/filesto list generated files,GET /api/files/{path}to return one file,GET /previewto serve the generated site, andGET /api/healthas the health check. - The README presents a validator that would check every file exists and parses before the system says “build complete,” evaluations measuring build success rate, time, and cost on fixed prompts, and automated tests for the file tools and validator as future work.
The README lists build history that would save past builds so they survive a restart as future work; it does not report measured build success, runtime, or cost.
I built Open Book, an AI-powered book discovery and question-answering application. It scrapes book information, stores it in a database, generates semantic embeddings, and supports AI-powered book summaries and question answering. Its stack includes Python, Django, Django REST Framework, React, Selenium, MySQL, ChromaDB, Sentence Transformers, OpenRouter, and GPT-4o-mini. I also built TableFlow, a QR-based restaurant ordering system using Python, Django REST Framework, React, and Railway; it provides digital menus and ordering workflows.
My other work includes a Django Blog Application, a URL Shortener, AI-powered document question-answering experiments, LangChain and LangGraph chatbot experiments, and web scraping and automation projects. Across my work I use Python, JavaScript, SQL, and Java (previously for DSA); Django, Django REST Framework, FastAPI, REST APIs, authentication and authorization, and multi-tenant architecture; and AI tools and methods including RAG, vector databases, embeddings, semantic search, prompt engineering, LLM integrations, agents, conversational memory, and tool calling.
My broader technical toolkit includes FAISS, ChromaDB, Sentence Transformers, PostgreSQL, MySQL, Redis, Celery, Docker, AWS EC2, Linux, Nginx, Gunicorn, Git, GitHub, React, HTML, CSS, and JavaScript. I have deployed full-stack applications on Linux servers and worked with Ubuntu and Kali Linux, Nginx reverse proxies, Gunicorn, HTTPS, domains and subdomains, GitHub-based deployments, PostgreSQL, and Redis and Celery. I maintain deployed projects and experiment with improving their performance, reliability, and architecture.
I prefer practical learning to purely theoretical study, and I like understanding how backend architecture, APIs, databases, and AI workflows work internally. I prefer complete applications over isolated demonstrations, maintainable code, realistic scope, and solving actual problems. I want to understand why and when a technology is useful rather than adopt it because it is trending. I value architectural understanding, testing, deployment, and direct feedback and honest technical criticism over generic encouragement; I prefer functional implementations over projects that look polished but lack substance.
I have covered LangChain fundamentals and basic RAG implementations. My current focus is shifting toward agentic applications, workflow orchestration, evaluation, and production-oriented AI engineering. My learning priorities also include advanced Python and backend engineering, Django and Django REST Framework, asynchronous FastAPI, multi-agent systems, AI agent memory and persistence, Model Context Protocol (MCP), PostgreSQL optimization, Redis and Celery, Docker and deployment, MLOps fundamentals, and scalable system design.
I practice DSA primarily in Python to improve my problem-solving skills and prepare for technical interviews. My experience includes arrays and strings, binary search, two pointers, linked lists, stacks and queues, and binary search trees; I’m continuing to strengthen graph algorithms, dynamic programming, backtracking, heaps, and advanced problem-solving patterns.
I’m looking for opportunities to work on real products, contribute to engineering teams, and gain experience building production-level software. I’m particularly interested in startups and product-focused companies working on AI applications and backend-heavy systems. My preferred work includes Python backend development, Django and FastAPI, AI-powered applications, LLM integrations, RAG pipelines, AI agents and automation, and backend infrastructure. Long term, I want to be able to independently design, build, deploy, and maintain scalable AI-powered applications.
I’m adding this professional-platform update to my profile alongside my work, projects, and career interests.
I joined ASX on 2026-11-11.
Looking forward to meeting folks on this platform to make meaningful connections.
This update records both my excitement about joining and my intention to build meaningful connections with people on the platform.
I am an AI Application Engineer focused on Agentic AI, RAG, and LLM applications. I am a final-year Computer Science Engineering student building AI-powered products, and I am looking for AI Engineering and Backend Engineering internships and entry-level roles where I can build scalable systems, APIs, and practical AI products.
My headline identifies my degree as B.Tech CSE ’27.
I'm a final-year Computer Science Engineering student building AI-powered products — agentic AI systems, RAG pipelines, and LLM applications that actually ship and scale.
My current position is Back End Developer at Self / Personal Projects (Jan 2025 to Present), Vadodara.
In this role, I build and deploy backend-focused Django applications, including production-style systems with authentication flows, database modeling, and REST API development. My backend work uses PostgreSQL, Redis, Celery, Docker, and AWS; I use Git and GitHub for version control and documentation, and practice Data Structures & Algorithms in Python daily.
My education is Parul University, Vadodara: Bachelor of Technology - BTech, Computer Science and Engineering (Aug 2023 to Aug 2027).
My high school education is YMS Sr Sec Public School: High School Diploma, CBSE (2022 to 2023).
I earned the certification - AI Fundamentals, Sep 2025, IBM SkillsBuild.
I earned the certification - Complete Data Science,Machine Learning,DL,NLP Bootcamp, Aug 2026, KRISHAI Technologies Private Limited.
I also practice Data Structures & Algorithms daily, with 150+ LeetCode problems solved, and I'm strengthening DBMS, Operating Systems, Computer Networks, OOP, and System Design.
My skills span agentic AI, LangGraph, LangChain, LangSmith, RAG, LLM applications, NLP, embeddings, semantic search, vector databases, FAISS, ChromaDB, Hugging Face, prompt engineering, classification, machine learning, artificial intelligence, and Scikit-learn. I also work with Python, Django, Django REST Framework, FastAPI, Flask, REST APIs, PostgreSQL, MySQL, MongoDB, SQL, Redis, Celery, database design, authentication and authorization, JWT, JSON, Express.js, and Node.js.
My infrastructure and development tools include Docker, AWS, Amazon EC2, Git, and GitHub. Other technologies and areas I list include JavaScript, TypeScript, C, HTML5, CSS, data structures, algorithms, OOP, problem solving, system design, DBMS, operating systems, computer networks, Kali Linux, and cybersecurity.
RAGBot is a multi-tenant B2B SaaS RAG platform for businesses to upload PDFs and chat with them through an embeddable widget. I built it with Django REST Framework, PostgreSQL, Celery, Redis, LangChain, FAISS, SentenceTransformers, and OpenAI GPT-4o-mini, and deployed it on AWS EC2 with a custom domain and HTTPS. It supports asynchronous document processing, contextual responses with source attribution, and isolated document processing per tenant. Its project dates are listed as Apr 2026 and Jan 2026.
• CineMatch — a content-based movie recommendation engine over 45,000+ movies, using NLP preprocessing, TF-IDF (50,000 features), cosine similarity, and FastAPI.
• Built a content-based recommendation engine using a dataset of 45,000+ movies.
• Used TF-IDF vectorization with unigram and bigram features and 50,000 maximum features.
• Generated a sparse matrix containing 1.7M+ non-zero entries and used cosine similarity for item-to-item recommendations.
CineMatch combines movie plot overview, genre, and tagline into a unified text corpus. Its NLP preprocessing includes lowercasing, regex cleaning, stop-word removal, and lemmatization; trained artifacts are serialized, and recommendations are served through an API. Project dates are listed as Sep 2026 and Sep 2026.
TableFlow is a QR-based restaurant ordering backend for menu browsing and order lifecycle management. It uses Django, Django REST Framework, PostgreSQL, and Docker, with JWT authentication and role-based access control for customer and admin roles. Its order states transition from pending to preparing to served. Project dates are listed as Sep 2025 and Aug 2025.