Thirugnanam V S
Software DeveloperVLSI EngineerAI ResearcherEmbedded Systems EnthusiastSoftware Developer

Parkinson's AI Monitoring

A machine-learning pipeline that detects Parkinsonian tremor from wearable IMU sensor streams. Built to research standard: subject-wise splits that eliminate temporal leakage, correct preprocessing order, and deep-learning architectures matched to tremor dynamics.

Data-Driven Control of Inverted Pendulum Systems Using Reinforcement Learning

Paper submitted to IEEE PIICON 2026

Benchmarks DDPG and PPO policies against a classical LQR controller for stabilising inverted pendulum systems, using a custom state-aware reward function. The accompanying paper has been submitted to the 2026 IEEE 12th Power India International Conference (PIICON).

Fake News Detection Using Machine Learning

End-to-end fake-news classifier built on the ISOT dataset with a modular Python architecture: a disciplined data pipeline, text preprocessing and feature extraction, model comparison, and all reported metrics derived strictly from the held-out test set. Fully documented, including an IEEE-format paper and project report.

Phishing Email Detection

Machine-learning system that classifies phishing emails from legitimate mail using engineered text and header features. Developed in structured phases with a modular codebase, rigorous train/test hygiene, and complete documentation from data pipeline to final report.

RS-485 Serial Communication between STM32F103C8T6 and Arduino UNO with Checksum Verification

Reliable board-to-board communication over an RS-485 differential bus between an STM32F103C8T6 and an Arduino UNO, with a checksum layer to detect corrupted frames. Covers transceiver wiring, frame design, and error handling on both microcontrollers.

Smart Task Scheduler

A JavaFX desktop application that schedules tasks through a comparator-driven priority queue, ordering work by priority level and deadline instead of a flat to-do list — so the most urgent items always surface first.

Automated Student Attendance System

A web-based attendance system on Google Apps Script that lets students mark presence from their own devices while the backend verifies authenticity with network-based checks and time-boxed sessions, writing verified records straight into structured sheets.

Job Portal

A Java job-portal backend built on JDBC and MySQL, implementing user registration and data operations through a cleanly layered DAO architecture — separating models, data access, and database connection handling.

Student Management CRUD

A student records management application implementing clean create, read, update, and delete operations over a relational database, with input validation and a clear separation between data access and interface layers.

TinyTapeout TT10 Shift Register Challenge

A shift-register design taken from RTL all the way to GDS through the TinyTapeout TT10 shuttle — including Verilog implementation, testbench verification, and a CI/CD pipeline on GitHub Actions that runs the full check-and-harden flow on every commit.

24-Hour Railway Clock

A 24-hour digital clock in the style of railway station displays, designed at the circuit level and built as a working breadboard prototype — covering counter design, display driving, and timing, from schematic to functioning hardware.

Automated Clothes Hanger System with Rain Detection and Shade Mechanism

First-year engineering hardware project

A first-year engineering hardware project: a clothes-hanger system that detects rain with a sensor module and automatically retracts the line under a shade mechanism, combining sensing, actuation, and simple control logic into a complete working build.