Publications
Research writing
Work from the academic side, where the standard of proof is somebody else trying to take the argument apart. Four papers, from two undergraduate hardware builds to a Springer journal article.
- 01
Survivability of Supply Chains in the Era of Industry 4.0
Piyal Sarkar, Angappa Gunasekaran, Harshwardhan Kiran Patil
Global Journal of Flexible Systems Management, 26(1), 225–246 · 2025
DOI 10.1007/s40171-024-00428-6
A study of what lets a supply chain survive disruption once Industry 4.0 technologies are part of it, published in Springer's Global Journal of Flexible Systems Management. My contribution was the analytics layer: the review pipeline across 200-plus papers, the coded datasets and evidence matrices behind the qualitative findings, and the grouping of challenges raised in 14 executive interviews at firms including Amazon, Nestlé, and Loblaws into the five analytics themes that became the paper's disruption response framework.
- 02
Time Series Stock Price Forecasting: An LSTM Neural Network Approach for Market Prediction
Harshwardhan Patil
Penn State ScholarSphere (master's culminating paper, M.S. Information Systems) · 2025
DOI 10.26207/be30-q097
My master's culminating paper at Penn State, and the piece of academic work closest to what I build now. It develops a three-layer LSTM architecture of 200, 200, and 64 units with tuned dropout regularisation, and pairs it with dynamic window normalisation and automated feature engineering to handle the parts of financial time series that break naive models: non-stationarity, regime shifts, and scale drift across a long history. Validation runs against Apple (AAPL) price data, testing whether the network holds up on short-term movement and longer trend at the same time, with risk quantification alongside the point forecasts rather than bolted on after. Deposited under CC BY 4.0 with a persistent DOI.
- 03
Design of Robotic Fish for Aquatic Environment Monitoring
Harshwardhan Patil
International Journal of Advances in Engineering and Management, 2(1), 74–79 · 2020
An undergraduate design paper on a robotic fish for detecting debris in open water, built around a constraint that shapes the whole system: it runs on batteries, so every computer-vision decision is also an energy decision. Waves make a stable camera view impossible, so frames are registered against the horizon line rather than the corner features that water environments do not provide, and debris is separated from background in HSV, which survives the reflections. A Raspberry Pi handles vision, an Arduino handles motion and the sensor array, and image processing is offloaded to the cloud or kept local depending on measured link speed. A rotation schedule trades movement energy against how much of the area stays covered.
- 04
Design and Making of Head Motion Controlled Wheelchair
Harshwardhan Patil
International Journal of Research in Engineering, Science and Management, 3(6), 80–84 · 2020
A wheelchair a quadriplegic user drives by tilting their head, designed around cost rather than capability. Powered chairs that a user with no hand control can operate exist; they are priced out of reach of most of the people who need them. This one replaces the microcontroller with tilt sensors feeding a diode logic circuit, four directions and a stop state built from analog parts, transmitted wirelessly from a hat to relays driving the geared motors. Built and tested with a rider, rated to 100 kg on a 1:10 gear reduction, and costed at roughly ₹30,000 against commercial alternatives.
The two 2020 papers are undergraduate engineering work at SRM: built hardware, not simulations. The later two come out of Penn State: a master’s paper of my own, and the Springer article where my contribution was the analytics layer behind the qualitative findings. The Experience page covers that research work in more detail.