Seminar on Interviewing Skills for Grad Students

Last Monday I had the privilege of organizing and hosting a seminar on "Interviewing Skills for Grad Students." It was a huge success and some students were even standing. We had 5 exceptional presenters from industry and academia. Seminar covered basic interviewing skills, behavioral interviewing process, success factors in interviewing and carrier, differences in start-ups and R&D teams, interviewing process in academia for faculty/PostDoc positions, etc. Based on the follow-up server, it seems that students benefited a lot from being able to talk and share experiences of the presenters.
Students from UMASS also joined the seminar over videoconferencing. All together we had ~40 students from both campuses. The seminar was sponsored by CASA Student Leadership Council (SLC). UMASS is a main partner of the CASA project. Generous support from all presenters, SLC leadership, and members was immensely helpful in making this event such a success.

Slowness, Unreliability, and Errors are Pervasive to Facebook

Facebook (FB) is already part of life for millions of individuals across the globe. It's already among the Internet giants like Google, Yahoo, YouTube, Amazon, & e-Bay. It has evolved as both a social interaction tool & a real-time communication tool (many Yahoo & MSN messenger users have switched to FB because of the chat feature). Unfortunately, most users will agree that FB is nowhere close the standards set by Google, YouTube, & Amazon when it comes to responsiveness, reliability, & availability. Personally, I have experienced extremely slow page loading, many error messages, not allowing me to login or post messages, and show only subset of postings by others. Hence, I don’t depend on FB to contact others & have a low appreciation as a useful service.

FB seems to be relying on extensive JavaScript & extended variants of MySQL, PHP, & Memcached. It's understandable that FB has to handle so much data, users, & their relationships. Unfortunately, heir underlying architecture seems to be not scalable. MySQL & PHP may be extended only to a certain extent. May be FB needs to look into completely new approaches like what Google (MapReduce, GFS) & Amazon (Dynamo) have done. Alternatively, many poorly written 3rd partly applications are breaking FB than making it. I don't have a solution for these problems, but I feel this is a good opportunity for research in handling large-scale data objects in a multi-user interaction based distributed environment.

Get Motivated by Tamara Lowe

After many years, I finally finished reading a nontechnical book.
Its "Get Motivated" by Tamara Lowe. Book is good & worth reading. According to her, motivation is based Drive, Needs, and Awards (what she called DNA). People have either Production or Connection drive, Stability or Variety needs, & Internal or External awards systems.
I think, I equally share all categories in DNA though the questionnaire in book categorized me into one of them. First few chapters were interesting & highly motivating but the middle was not. Nevertheless, as she said in the last chapter "Finish First, Finish Well" the book ended really well.
To be honest, I feel renewed and motivated than in the recent past.

Why model Packet Arrivals as a Poisson Process?

In queuing theory-based analysis, we always assume arrivals to be according to a Poisson process. "Why we do so, why not it's uniform or some other distribution?" This was a question raised by my supervisor & a colleague while I was preparing for my Qualifying Exam. As I was not able to answer well, my supervisor gave some hints. I was worried that the same question may come up during the exam so searched for an answer on web. Unfortunately, no specific answer was found. So I'm trying to lay down an answer by adding bits & pieces from here & there & my supervisor's answer.

It can be answered by looking at the properties of a Poisson Process. Recall that Poisson Processes are used to model statistically rare events.

A counting process {N_t, t ≥ 0} is a Poisson process if:

1. N₀ = 0
2. N_t has stationary independent increments
• N_t1 - N_s1 is independent from N_t2 - N_s2
• Memoryless
• Inter arrival times are independently & identically distributed set of exponentially distributed random variables
3. P{N_Δt = 1} = λ(Δt) + o(Δt)
4. P{N_Δt = 2} = o(Δt)

Property 2 (memoryless) is the key. With uniform distribution if we say, "probability of an arrival of a packet is 0.1," if a packet doesn't arrive between [0 - 9] it has to arrive at t = 10. In a natural system, such arrivals are not necessary. A packet may or may not arrive or if an Earthquake doesn't happen in Yellowstone National Park for last 9 years it doesn't necessary mean it must occur in the coming year. Hence this reflect the natural behavior & makes the analysis easier.
Property 3 & 4 says that there will only be one arrival at a time (no batch arrivals). This holds for a small time interval Δt → 0. Hence, we only need to worry about 1 arrival at a time. Even if 2 Earthquakes occur during 5 minutes, it will happen one after another. This further simplify the analysis.
Given these properties Poisson is a better & simple approximation.

Please feel free to suggest any corrections or additions....

Failed - But not forever...

We fail in many endeavors...
But never quit trying...
You may succeed either in the same or different area...
Even if you succeed in a different area, it because of the previous failure....
Here are some well-known failures that could inspire you...

Why Linux Sucks! - Reality behind some issues in Linux

An interesting presentation by Bryan Lunduke that talks about some of the realities behind Linux & it's future. I do agree that Linux needs a sustainable economic model. All the smart/nerdy developers can contribute while they are teens or while in college. But when you step into real life, you need some serious income. If there is a model that has some economic motives for those developers, we'll have tremendously better software & an OS while enabling those developers to do what they love to do...
Slides are available at Bryan's web page.

A United Sri Lanka...

Finally, the Sri Lankan war is over & everyone is in a single country...
I remember the impact of war as far as I remember anything in my childhood... I still remember how windows got scattered & fall down while I was in school due to the JOB bomb (1991). Finally, being able to hear that war is over is a dream come true...
Now it's our responsibility to the rebuild the country & make sure people in those areas are taken well care of.
Salute to the solders for all forms of sacrifices & president for the leadership...!