AI Researcher & Deep Learning Developer passionate about building intelligent systems using Computer Vision, Multimodal AI, and Deep Learning.
Complete, self-contained firmware for a six-legged robot with three joints per leg. It drives 18 hobby servos through two PCA9685 boards, performs three-link inverse kinematics, generates a non-blocking tripod gait, stores servo trims in EEPROM, and exposes a safe serial command interface.
AI-powered video analytics platform: vehicle detection, multi-object tracking, perspective localization, speed & spacing analysis, convoy/group detection, event engine, and reporting — behind a dark operations dashboard.
The AI-Powered Disaster Response Mesh Network is an intelligent emergency communication and survivor detection system built using ESP32-S3, LoRa, TinyML, and environmental sensors.
A complete local research application for civilian UAV observation: React/TypeScript control-room UI, Python/FastAPI vision service, YOLO inference, ByteTrack, motion analysis, telemetry provenance, session recording, replay, and exports.
this is a safe, general research alternative to the requested military-person detection system. It does **not** implement a soldier detector, specialized camouflage optimization, ACD-Net hybrid fusion, autonomous targeting, identity tracking or a validated real-world detector. The included trained checkpoint recognizes synthetic geometric shapes.
This package is the runnable first implementation of the supplied master prompt. It provides a local monitoring dashboard, camera-source abstraction, face detection, temporary multi-object track IDs, manual target locking, target-loss state, live telemetry, and research-session recording. The default mode is simulation/observation.
The app stores roughly one metric sample per source second. Event onsets are saved at processing time. Pauses and missing frames do not contribute unobserved dwell time. Seek is disabled during recording. Source changes finish the current recording first. The initial sample is saved immediately in image mode. Source video is never
A working local computer-vision workstation based on the supplied dashboard reference. React + strict TypeScript + MediaPipe + Three.js, with an optional FastAPI / SQLite API. It includes the source, local model files, a prebuilt frontend, Windows/macOS/Linux launchers, tests and documentation.
FaceScope recreates the observable capabilities in the supplied demonstration: a dense face mesh, iris markers, eye close-ups, facial motion coefficients, and head orientation. It is an original implementation using MediaPipe, not the proprietary software shown in the recording.
A modular desktop workbench for exploring face motion in webcams, images, videos and supported RTSP/IP streams. Select a face, lock its temporary track, inspect timestamp-based motion and export the measurements. Frames are processed locally; recording is off until explicitly started.
Smart Mushroom Farming System using ESP32 + IoT + Computer Vision An AI-powered IoT-based Smart Mushroom Farming System designed to automate and optimize mushroom cultivation using: ESP32 IoT Sensors Relay Automation Computer Vision Cloud Monitoring AI Disease Detection
An interactive, locally hosted mathematics laboratory by Shivam Singh / MathTech. It uses a small browser interface, a Python HTTP API, NumPy, SciPy and SymPy. No account or hosted service is required.
The capstone models future churn from eight current customer snapshot features. It validates input, removes duplicate evidence, creates independent train/validation/test partitions, fits preprocessing within training folds, compares models, searches hyperparameters, selects a decision threshold on validation data, and evaluates the frozen procedure
Understand the mathematics → implement → visualize → experiment → apply → build.
**A working, layered C17 TCP client–server application.** Prepared for **Shivam Singh · MathTech**. Linux / POSIX, GCC, Make, pthreads. No database server, third-party C libraries, Python, Node.js, or package download is needed to build or run the application or its tests.
This project is an advanced 5-axis AI-powered robotic arm system built using: ESP32 Arduino Mega PCA9685 Servo Driver Computer Vision AI Learning Models Real-Time Motion Planning
A Python numerical-analysis textbook and computational laboratory, spanning floating-point foundations, scalar roots, matrix methods, approximation, calculus, ODEs, PDEs, eigenvalues, optimization and scientific applications. Manual implementations are paired with independent analytical or scientific-library checks.
A production-oriented **FastAPI modular monolith** for hospital operations. It combines patient administration, clinical workflows, pharmacy, laboratory, inpatient care, billing, insurance, analytics and auditing in one runnable backend that can later be decomposed into services.
A production-oriented **modular monolith** backend for a food-delivery platform. It contains complete workflows for customers, restaurant owners, administrators and delivery partners and is structured so modules can later be separated into microservices.
A production-oriented FastAPI backend for direct messaging, groups, channels and real-time collaboration. The project combines a full identity service with REST APIs, WebSockets, Redis fan-out, PostgreSQL persistence, moderation, notifications, audit logs and deployment assets.
A production-oriented FastAPI backend for multi-property hotels, resorts and serviced apartments. It is implemented as a modular monolith that can later be split into identity, reservation, property inventory, front-desk, operations, billing and analytics services.
A production-oriented FastAPI backend for multi-category online retail. The project is implemented as a modular monolith that can later be split into identity, catalog, inventory, order, payment, procurement and fulfilment services.
A standalone identity and access management backend for web, mobile, SaaS and internal enterprise applications. The project is designed as a production-oriented modular monolith that can later be extracted into a dedicated identity microservice.
The learning sequence is: define the mathematical problem, derive a finite approximation, implement it transparently, visualize its geometry, measure error and cost, investigate failure cases, and apply it to a scientific model. Reusable algorithms live in `src/`; SciPy and mpmath provide separate validation references
A reproducible Python research project for the centered multivariate exponential-power family: analytical proof, exact sampling, identifiable estimation, repeated experiments, 13 notebooks, publication figures and a generated manuscript
A professional 6 Degrees of Freedom (6-DOF) Robotic Arm built using Arduino Mega 2560, high-torque servo motors, and Bluetooth communication.
This repository combines a compact course, reusable Python algorithms, numerical experiments, 2D/3D figures, animated demonstrations, and application pipelines. It starts with vectors and builds toward SVD, PCA, numerical methods, machine learning, image geometry, robot kinematics and camera projection.
A modular ESP32 firmware project for passive 2.4 GHz RF activity monitoring using an **nRF24L01+**, a **16x2 I2C LCD**, and a **single menu push-button**. The firmware provides a responsive, non-blocking four-mode menu for observing RF activity across selected portions of the 2.4 GHz ISM band.
A guided computational course from vectors and multivariable limits to optimization, multiple integrals, vector fields, Green’s theorem, Stokes’ theorem and the divergence theorem. Each lesson includes a worked derivation, symbolic computation, numerical verification, complementary 2D/3D views, a widget experiment and 16 exercises.
`single-variable-calculus-lab` is a computational mathematics laboratory for students, educators, engineers, and independent learners. It connects each major idea in single-variable calculus to symbolic algebra, numerical experiments, reusable Python, clear 2D/3D graphics, interactive widgets, and applied projects.