Dhiraj Reddy — AI / Software Engineer

Computer Science undergraduate building practical AI systems, backend applications and developer tools. Based in Hyderabad, Telangana, India. 4th-year B.Tech Computer Science undergraduate.

Featured and selected projects

• Verascope — Ask a codebase questions. Agents investigate bugs with cited evidence, test fixes in a Docker sandbox, and wait for your approval. Status: LOCAL ONLY. Stack: Python, FastAPI, PostgreSQL, SQLAlchemy, ChromaDB, React, TypeScript, Docker, Ollama.
• ScoutLens — Investigate a job or internship before you invest your time — live SerpApi searches, an evidence store, and a report where every claim links to a real source. Status: DEMO. Stack: Python, Streamlit, Pydantic v2, SerpApi, SQLite, BeautifulSoup, Ollama (optional).
• SCARFLOW — Supplier corrective-action (SCAR) and 8D workflow for manufacturing quality teams — replacing Excel sheets and email threads. Status: LOCAL ONLY. Stack: Next.js 16, React 19, TypeScript, Tailwind CSS 4, Supabase (Postgres + RLS), Inngest, Resend, Playwright.
• SmartRide AI — Edge-AI rider-safety assistant for two-wheelers: a camera watches for a missing helmet, phone use and hands off the handlebar, and raises an immediate alert. Status: HARDWARE. Stack: Python, Raspberry Pi, YOLOv8, TensorFlow Lite, OpenCV, MediaPipe, MPU6050.
• RainFogHaze — Haze and fog removal with OpenCV's dark-channel prior, where a local vision model (LLaVA) chooses the dehazing strength. Status: DEMO. Stack: Python, Django, OpenCV, NumPy, Ollama, LLaVA.
• Study Buddy — A study assistant that answers only from the PDFs you uploaded for the selected subject. Status: LOCAL ONLY. Stack: Python, Streamlit, LangChain, ChromaDB, Ollama, llama3.1:8b.
• Fake News Predictor — Paste a news-article URL; a TF-IDF + logistic-regression model trained on an Indian fake-news dataset labels it FAKE or REAL with a confidence score. Status: DEMO. Stack: Python, scikit-learn, TF-IDF, Newspaper3k, Streamlit.
• AI Interview Coach — A fully local interview-prep chatbot: Streamlit + LangChain + Ollama running DeepSeek-R1:8B, with four answer styles and conversation memory. Status: LOCAL ONLY. Stack: Python, Streamlit, LangChain, Ollama, DeepSeek-R1:8B.
• Job Application Tracker — Full-stack tracker for job applications: FastAPI + SQLAlchemy + MySQL behind a Streamlit dashboard, with an end-to-end API test script. Status: LOCAL ONLY. Stack: Python, FastAPI, SQLAlchemy, MySQL, Pydantic, Streamlit.
• AI Analytics — Upload a CSV or Excel file and ask questions: a LangChain agent answers by calling 12 tools for statistics and charts, with Qwen3 running locally. Status: LOCAL ONLY. Stack: Python, Streamlit, LangChain, Pandas, Plotly, Ollama (qwen3:4b).

Other projects

• Student Management System — Add, view and delete students — a web app in plain Java (JDK HttpServer + JDBC), MySQL, HTML and CSS with no frameworks.
• Scraper Chatbot — Streamlit scraper that extracts a page's title, same-site links and text. The RAG modules (chromadb / embedding / retrieval) are empty placeholders, so there is no chatbot yet.
• PDF Reader (Ollama) — Streamlit chatbot that answers questions from an uploaded PDF using pypdf and DeepSeek-R1:8B through LangChain-Ollama.
• Movie Recommender — TMDB 5000 movie recommender: notebook builds the similarity assets, a Streamlit app serves recommendations.
• Iris Predictor — Streamlit app that predicts iris species from four measurements using a pickled model and label encoder.
• House Price Prediction (XGBoost) — Kaggle “House Prices” workflow: EDA, preprocessing, feature engineering and an XGBoost regressor with cross-validation.
• Spam Mail Classifier — Text-classification notebook on a Kaggle email-spam dataset with vectorisation and classical models.
• Heart Disease Predictor — End-to-end ML notebook on the UCI heart-disease data: imputation, scaling, encoding, model training and evaluation.
• Titanic Survival Analysis — EDA, feature engineering and survival prediction, ending in a YData Profiling HTML report.
• Netflix Content EDA — Exploratory analysis of the Netflix catalogue: release year vs year added, movies vs TV shows, genres.
• House Price · Linear Regression — Task 01 of the SkillCraft Technology internship program: linear regression on square footage, bedrooms and bathrooms.
• Linear Regression from Scratch — Gradient-descent linear regression derived by hand and implemented with NumPy only, with a fitted-line plot.
• Logistic Regression from Scratch — Logistic regression implemented from the formulas up and tested on the diabetes dataset.
• Smart Insights of Customer Patterns — A customer-segmentation project: the README describes Gender vs Spending Score analysis and clustering with feature engineering and the elbow method. The notebook and dataset currently in the repository are the heart-disease ones, so the clustering notebook is not published there yet.
• GenAI Basics — Notebooks on calling LLMs and on prompting types.

Sections

• RUN MY PROJECTS.exe — Project Playground — try the demos
• Projects.exe — Featured, built and important projects
• About Me.exe — Who I am and what I build
• Skills.exe — Skills, with the projects that prove them
• Resume.exe — View or download my resume
• GitHub.exe — Repositories, languages and recent activity
• Contact.exe — Email, GitHub, LinkedIn
• Terminal — C:\DHIRAJ — A portfolio terminal: try help, neofetch
• System Info — System information and what I'm exploring
• DHIRAJ.LOG — My blog, embedded
• STACK.exe — How my technologies relate
• JOURNEY.exe — Timeline of projects and milestones
• HOW I BUILD.exe — My build workflow
• Welcome.exe — Start here
• Built Projects.exe — Projects I engineered
• Important Projects.exe — Projects that matter to me, including AI-assisted ones
• Recycle Bin — Nothing to see here. Probably.

Contact

Email iamdhirajreddy@gmail.com. GitHub https://github.com/uppadadhiraj. LinkedIn https://www.linkedin.com/in/venkata-dhiraj-reddy-uppada-277bb828a/.