SAM's Playground

ACADEMIC PROJECTS

Error Analysis while performing Sequence Labelling for NLP Tasks
[Machine Learning - 2018]

Link: Sequence Labelling and error analysis.

• vsmlib to process word embedding models.
• Acheived 90% accuracy using sklearn multinomail Logistic Regression classifier.
• Achieved 97% accuracy with sklearn MLP neural network classifier
• Dataset: CONLL-2003 Shared Task English
• Lead the team and strived for active participation from them, helped them understand the underlying concepts

Travel Buddy App – [Internet and Web Systems - 2017]

Link: Travel Buddy App

• Used MEAN stack, Mongoose for schema definition,
• Deployed successfully on Heroku, using nmp start in dyno. Web-Scaling: 1
• Created upwards of 9 components, following MVC design patterns and RESTful APIs,
• • Components Include:
  • Home
  • Social Login
  • Trips management
  • Profile
  • Buddies
  • About us
  • Meet the Team
• Lead the team and strived for active participation from them,

THREEjs Demos – [Computer Graphics I - 2017]

Link: Computer Graphics Assignments/THREEjs Demos/jQuery Interactive

• Created various graphical representations for various assignments and finals.
• Used THREEjs, MVC design pattern, gave the user a complete set of buttons to work with views
• Industry standard code with ample documentation was provided.

RADAR-IMU system for 3D-DIC measurements using remotely paired UAVs – [FALL 2017 – SPRING 2018]

Link: IMU-RADAR system for 3D-DIC

• Ran experiments to calibrate a system composing of RADAR-FMCWR3400W/04, IMU-ICM 20948-STM 32 Nucleo64 interface.
• Used pySerial and pandas to receive data from RADAR/IMU, plotting through mathplotlib, data processing using numpy.
• Created interfacing software to include a RaspberryPI at the center of the system as a processing unit.
• Packaged with self-generated user manual.

Barcode Item Scanner System – [FALL 2016]

Link: py+MQTT - Barcode Item Scanner System

• Scanning barcode through camera, encrypted through MQTT authorization
• Developed a secure item scanner system, using Raspberry Pi and secured the data transmission using Mosquitto MQTT broker with certificate-based authentication.
• I learnt to create MQTT based authentication for a IOT network. I learnt the use of certificates.

Contra Game AI - Created an AI agent for a level of the game Contra – [FALL 2016]

Coming soon

• Created a simple agent to automate the gameplay for a game level developed by some external resource, using C++ and their game engine.
• Concept: Simple Reflex agents/ Evaluation function etc.

Prime number generator – Simple Machine learning project – [SPRING 2018]

Coming soon

• Created a prime number checker and generator using Support Vector Machines to train.
• Used 100 consequent numbers to train and then used a validation set to check accuracy which was at 91%.

Image Based Passwords – [Capstone 2015]

Coming soon

• New type of authentication using Captcha as graphic passwords
• Three types of authentication, as a picture password, jigsaw puzzle, or button based.
• Used Bootstrap, WAMP and Dreamweaver, HTML, CSS, JS, PHP.
• Captcha as Graphical Password research paper was used as a base design pattern reference.

Flight Database Management System – [2013]

• Flight booking application.
• It was a fully functional management app, with GUI and user input, lots of tabs, easy to use.
• Used VB6, ADODB, SQL.

Smokes Counter – [Personal]

Coming soon

• Built android app to monitor smoking history. Android/Web TimeLine: Ongoing.