|
Contents:
1. Projects
2.
Patents
3.
Industry Related Projects
i. Yahoo!
ii. Microsoft
|
Projects
|
|
Area
|
Title
|
Resources
|
|
Mobile Computing
/
Distributed Systems
/
Spatial Indexing
|
Spatial
Alarms
|
Report
|
Sources
|
Simulator
|
|
Cooperative
Caching on Mobile Systems
|
Report
|
Sources
|
API
|
|
Comparison
of mobile caching algorithms
|
Report
|
|
Comparison
of spatiotemporal indexing techniques
|
Report
|
|
Caching
on mobile device
[Ongoing
Research with Prof. Ling Liu – submitted
to ICDCS’07]
|
|
Buffered
R Trees
|
Report
|
Sources
|
|
Visualizing
Tiger Data
|
Sources
|
|
Operating Systems
|
User
Mode Thread Library
|
Report
|
Sources
|
|
Barrier
Synchronization
|
Report
|
Sources
|
|
Windows
Serviceability (NTFS and ntoskrnl.exe) – as a Software Design Engineer at
Microsoft (1.5 years)
|
|
Machine Learning
&
IR
|
Generic
Recommendation Engine (as part of internship at Yahoo! Inc)
|
|
Rook
Isolation Bot
|
Report
|
Sources
|
|
Logic
|
|
Sources
|
|
Bayesian
Networks
|
Report
|
Sources1, Source2
|
|
Neural
Networks
|
Report
|
Report2
|
|
Health
Insurance Fraud Detector (Using Decision trees)
|
Report
|
|
Graphics
|
Interactive
Design of Triangular Meshes
|
Report
|
Sources
|
Demo
|
|
eCommerce
|
Extensions
to Auction House Application (AHA) with Prof. Y Narahari
[Won
the best project award at Computer Society of India’s *.fest, 2002]
|
2. Patent:
1. Submitted
patent pre-disclosure document for “Top N messy binaries – an adaptive
servicing model” while at Microsoft - Jan 2005
2. Submitted 2
invention disclosures on “Attribute based recommendations”
while at Yahoo! - Aug 2006
3. Industry
Related Projects:
i. Projects at Yahoo!
a. “DreamCatcher” - Generic Personalized Recommendation
Engine
Given the vast amount of information
available for customers it is necessary to build a system that could
understand customer needs and present the right set of items the customer
might be most likely interested in. The system was designed for generating
high quality results with focus on scalability, extensibility and
configurability. The task involved extensive literature review in the field
of personalized recommendations, Collaborative Filtering techniques,
Association rules and other data mining constructs. I also designed and built
a skeleton framework into which new algorithms could be easily plugged in
and evaluated. The framework was implemented in C++, Java and perl. The
work also involved implementing item based Collaborative Filtering techniques
and exploration into the field of ‘Attribute based recommendations’. The work
also resulted in submission of 2 invention disclosures at the end of the
internship.
b. Affinity Engine Algorithm for Y! Autos
- A Scalable Probabilistic Approach for matching cars
The
marketing advantage is very strong if customers are shown advertisements
that they are most likely to be interested in. This effort was to build a
system that would show consumers relevant links based on the
"affinity" between entities. The system is adaptive and improves
itself over time. The system is currently being used to show cars to users
that they are most likely to prefer based on their current navigation
history and past information gained from other users.
ii. Work at Microsoft:
At Microsoft I
was involved as a developer in the Windows Serviceability Team (WinSE) for
about 1.5 years. I was responsible for the windows base components namely –
NT kernel, NTFS and storage device drivers for Windows 2000 and NT4.
During my stay there I was
involved in developing the following:
a.) Hotfixes for
Win2000,XP and W2k3
b.) Critical
security patches released as GDR for several components for Win2000 and NT4
c.) Win2000 update
rollup package fixes
d.) Design change
requests
e.) Request for
collaboration (basically helping out PSS folks by answering their
questions)
Most of my work
revolved around debugging & fixing issues by looking at memory dumps,
remotes and customer scenarios.
BTW, if you are running Win2000 URP+,
Win XP SP2+ or Windows 2003 SP1+ I am happy for obvious reasons! :-)
|
