ahlawat.net

Current Work

I work at Desktone a startup in the desktop virtualization space. Desktone is a small team of engineers - all of whom are super bright, professional and a lot of fun to work with. I am currently writing the support for Xen hypervisor using the XAPI.

At HP

My latest project at HP was working on writing and researching the next generation of our product. I wrote most of the platform on which the distribution happened - spring, hibernate, mysql, JSW etc. Before that I was working in the UI team - where I introduced ajax over JSF using prototype and also introduced other technologies like Spring DWR, flex etc.

Here are a few memorable projects outside of work.

Robocup 2003-2004

I was in the Electrical Engineering team and was responsible for the development of the RF wireless link between the robots and the AI computer. Was very hard work and the learning curve was phenomenal. I did a lot of stuff from programming PICs to designing and populating a wireless board. In the end I also worked on a portable wireless signal strength detection tool which would dump the percentage of lost packets to a LCD. Good fun.

Here is a link to my masters thesis at Cornell.

Trishul - autonomous self navigating robot

I think Bruce Land is the best Professor I have ever had. The man is fun, energetic and an overall rock star. The microcontrollers class offered at Cornell is a showcase of some of the most creative projects in embedded engineering you will see from Atmel based TV games to IR X-Y scanners and embedded HTTP servers. I decided to build a autonomous navigation robot with memory based AI and real time obstacle detection. In the limitations of time I think we did a decent job (I aced the class!!) - of course it was more fun than any other academic experience I have had.

Trisul

KWheel.com

This is actually Paul English’s idea, I met him when I was interviewing at Kayak and both of got together informally in our spare time to put up a toy website called KWheel.com. The idea is very simple - the user uploads a picture and the server analyzes the image - and tells the users what colors occur the most in the image and what colors have the maximum amount of independent areas. I have coded this in a mix of groovy/grails and java and use image quantization followed by a recursive pixel fill algorithm all java 2D and java image API. I coded most of it over a weekend - and put it up, its not production quality just a proof of concept.

A sample output from Kwheel

Really fun project. Would work on a V2 someday.