Manu Kothari

Mentor: Dr. Kavi Mahesh, Dean of Research, PES University, Bangalore

EduVideo is a video-sharing platform facilitating knowledge exchange through short, concise and crowdsourced video lectures (of length less than 3 min to maintain modularity) specific to curriculum/syllabus with the best of features of Youtube recreated on a smaller scale with an intention to eliminate the need for hand-written and photocopied notes and ensure uniformity in education

• Recommending videos – LSH… Show more

Mentor: Dr. Kavi Mahesh, Dean of Research, PES University, Bangalore

EduVideo is a video-sharing platform facilitating knowledge exchange through short, concise and crowdsourced video lectures (of length less than 3 min to maintain modularity) specific to curriculum/syllabus with the best of features of Youtube recreated on a smaller scale with an intention to eliminate the need for hand-written and photocopied notes and ensure uniformity in education

• Recommending videos – LSH (Locality Sensitive Hashing) by vectorizing user behavior and hashing it into buckets depicting similar users
• Personalized Search – Ranking semantically extracted information after mapping user query to video collection using TFIDF
• MongoDB | GridFS | Locality Sensitive Hashing (LSH) | Python (Flask) | Bootstrap | PHP | NLP | Apache Benchmark
Show less

Mentor: Prof. N S Kumar (Visiting Professor, PESIT)
• With an aim to make polyglot programming simpler, I worked closely on developing a generic and intelligent framework using which python scripts can be extended from and embedded into C/C++ with just 2 lines of code by abstracting away the low-level data structures and memory handling required for internal communication.
• C++ | Python | Linux

Mentor: Prof. Phalachandra H L (Cloud Computing and Big Data, PESIT) [email protected]

• As a part of our Software Engineering (SE) Project, I played the role of Team Lead, where we developed a software to analyze events which might have led to inflection points in the stock value curve of a given company.
• It also predicted the stock values for the next few days using the previous patterns of the stock curve and we achieved an accuracy of 81.3%

Mentor: Prof. Nagegowda K. S. (Embedded Systems Technology Domain, PESIT)
• Improved voice recognition by 26% by nullifying effect of environment noise using edit distance
• Transferred data between server/client using python sockets
• Raspbian OS | Python (with PySpeech) | Socket Programming

Mentor: Prof. Nitin V Pujari, HOD CSE, PESIT
• Anomalous behaviour was graphically measured by the variation of white pixels in the foreground mask of each frame in the video
• A dynamic threshold learning algorithm was employed to offset the environmental noise, thereby, improving accuracy to 79.8%
• Devised a disk-save algorithm to detect similar and redundant suspected frames which saved 88.12% secondary storage space compared to our previous solution
• C++ | OpenCV | Algorithms… Show more

Mentor: Prof. Nitin V Pujari, HOD CSE, PESIT
• Anomalous behaviour was graphically measured by the variation of white pixels in the foreground mask of each frame in the video
• A dynamic threshold learning algorithm was employed to offset the environmental noise, thereby, improving accuracy to 79.8%
• Devised a disk-save algorithm to detect similar and redundant suspected frames which saved 88.12% secondary storage space compared to our previous solution
• C++ | OpenCV | Algorithms | Computer Vision | Linux
Show less

Mentor: Prof. H A Dinesha (Crucible of Research and Innovation (CoRI), PESIT)
• Decreased probability of hacking to 2E-11 for a 5-Dimensional Password.
• Java | Cryptography and Security | Linux