Machine Learning and full-stack developer building ML systems, RAG agents, and web applications. 23 public repositories and counting.
I'm a developer working at the intersection of machine learning and full-stack engineering.
- Currently building: ML models, RAG pipelines, and full-stack apps
- Focus areas: NLP, document understanding, time-series forecasting, LLM agents
- Core stack: Python (ML, Streamlit, LangChain) + JavaScript/React (Vite, MUI, Tailwind) + SQL
- 23 public repositories spanning ML, NLP, RAG, and web projects
- Philosophy: ship working systems, document them well, iterate fast
I work on the full lifecycle — data prep and modeling, backend APIs, and usable frontends — with an emphasis on reproducible, production-minded projects.
| Project | What it does | Stack |
|---|---|---|
| Transformer-Based-Handwritten-Mathematical-Expression-Recognition-and-LaTeX-Transcription-System | Handwritten math expression recognition with LaTeX transcription | Python, Transformer, Deep Learning |
| S-P-500-Stock-Prediction-ARIMA-LSTM-Streamlit- | S&P 500 forecasting comparing ARIMA and LSTM with interactive charts | Python, Streamlit, Time-Series |
| Huggingface-langchain-ai-agentic- | Agentic AI workflows built on Hugging Face models and LangChain | Python, LangChain, Hugging Face |
| sql-fundamentals-rag-agent | RAG assistant for learning and querying SQL fundamentals | Python, RAG, LLMs |
| Document-Query-Assistant | Ask questions over your documents with retrieval-grounded answers | Python, RAG, NLP |
| Hostel-Management-System-Project | Hostel management web app with room and resident workflows | React, Vite, MUI |
More to explore:
- Movie-Genre-Classification — text classification for movie genres
- SPAM-SMS-DETECTION — SMS spam classifier
- AI-Based-Doubt-Resolution-Bot- — Q&A bot for resolving doubts
- DBMS_PROJECT — Streamlit + MySQL auction system
- Cisco_Packet_Tracercsma-cd-wlan-80211_MCP — networking / Packet Tracer work with MCP
- ML classifiers and predictors — genre classification, spam detection, stock forecasting with ARIMA/LSTM
- Document intelligence — handwritten expression recognition, document Q&A, LaTeX transcription
- RAG and LLM agents — LangChain pipelines over Hugging Face, Ollama, llama.cpp, and Gemini models
- Full-stack apps — React + Vite + MUI frontends backed by Python and MySQL
- Interactive demos — Streamlit dashboards that make models usable, not just notebooks
- Advanced RAG patterns: reranking, hybrid retrieval, evaluation harnesses
- Agentic workflows with LangChain and Hugging Face tooling
- Production ML: Dockerized inference, model monitoring, reproducible pipelines
- Stronger React architecture: Vite performance, MUI theming, Tailwind systems
- GitHub: github.com/SHYAMFRANCIS
- Open an issue on any repo to reach me — I read every one.
Thanks for stopping by — star a repo if something is useful.