Indian IT Industry

ManSingh · Post by **ManSingh** » 11 May 2017 00:15

Hmm interesting.
The stated purpose of GA TECH going online is to increase STEM candidates within usa. Although anyone can take their course from anywhere in the world after credentials have been validated by them. No land required. Funding might be needed for deserving candidates though.
I agree with your alliances portion. We need educational institutions, govt, banks and employers to collabrate and be willing to take risks( even failure). Currently all focus on their own bottomlines.
But first it requires a will to change..

Singha · Post by **Singha** » 11 May 2017 05:59

Arent foreign univ offshore campus banned in india for trumped up reasons like they would poach faculty from govt institutes? atleast the money would remain india and any work done benefit the local industry and economy than proudly proclaiming "oh india sends 200,000 BS+MS students to USA annually beating china by 15000"

thats all people and money feeding the western economy

Good point about curriculm development always trailing by 5 to 10 years in india.

I am all for inviting whoever wants to setup education here to come. middle rung univs may be interested and thats also a net gain never a loss . market will decide wages and whether its worth it to study there. any fraud univs will quickly get frozen out by the job market and slink away

we do have some peculiar notions of swadeshi.....swadeshi stuff(human resources) that can benefit from bideshi is not encouraged in favour of importing the Products & Services wholesale from bideshi often the creation of swadeshi people that went bideshi to rub more salt into the wound.

lack of enough scale and quality in higher education at value for money price point to anywhere near a world std is one of the single biggest roadblocks to india climbing up the economical curve.

Neshant · Post by **Neshant** » 11 May 2017 06:23

If they really want high tech design & manufacturing to take off in India, they *must* ensure quick and duty free imports of electronic components.

Failure to do this means there will be no benefit whatsoever from all the "make in India" screw-driver giri projects being setup as offsets after purchasing foreign military & civil goods at enormous expense.

Unless someone gets the ear of the PM and forces him to set up an electronics park which runs smoothly, there is no hope of India emerging as a design & development center in the 2nd wave of industrialization.

Prasad · Post by **Prasad** » 11 May 2017 10:20

While former ada and drdo heads decry the fact that components are imported easily and local players can't compete against a product made of tax free imported components.

Vadivel · Post by **Vadivel** » 11 May 2017 11:55

Is anyone here in the ML/AI field?

Like to get some advise on JV? please ping me at mk at rhytha dot com

hanumadu · Post by **hanumadu** » 11 May 2017 22:40

Has anybody seen AI in action in the IT field? IBM Watson or anything anywhere? What kind of jobs is it supposed to do and replace? When is the expected time of deployment in large scale across the industry?

IBM Watson is not particularly good in the one place I have seen. Predicting cricket scores.

vivek_v · Post by **vivek_v** » 12 May 2017 06:44

hanumadu wrote:Has anybody seen AI in action in the IT field? IBM Watson or anything anywhere? What kind of jobs is it supposed to do and replace? When is the expected time of deployment in large scale across the industry?

IBM Watson is not particularly good in the one place I have seen. Predicting cricket scores.

I have been working in the "AI" field for the past 4.0-4.5 years though its not AI but in Machine learning and Deep learning.

I primarily use it in Imaging and in Signal processing where it is really helpful in creating some unique designs which were not previously possible or were extremely time-consuming to create. There are other organizations who use in Speech combined with NLP (Alexa, Siri..etc). Also, it has been proven very successful in a lot of text-based jobs and works like in HFT (High-frequency trading), legal...etc. Of course, there are the usual use cases like recommender systems and other stuff which you can easily google.

If you ask me whether these techniques can replace programming jobs, I would say "no". The people/reporters tend to confuse ML techniques with automation and merge both since sounds nicer when they type the articles. The "automation" which is taking over is basically to write scripts and batch jobs to replace manual testing, sysadmin works in data centers and in certain other areas.

What actually seems to cause the slowdown in Indian IT is the proliferation of FOSS software and tools with the requirement for engineers who have programmed and deployed systems using these tools with minimal resources, comfortable in multiple programming languages and have experience in cloud deployments rather than people management skills.

I feel that the plight of organizations who supply expensive hardware with their software or vice-versa (leaving niche stuff like Teradata) would be even more terrible which is causing a chain reaction in the ODC's for those vendors run by Indian IT shops. No one nowadays wants an upfront huge Capex investment either in HW or in SW.

hanumadu · Post by **hanumadu** » 12 May 2017 08:56

vivek_v wrote:
hanumadu wrote:Has anybody seen AI in action in the IT field? IBM Watson or anything anywhere? What kind of jobs is it supposed to do and replace? When is the expected time of deployment in large scale across the industry?

IBM Watson is not particularly good in the one place I have seen. Predicting cricket scores.
If you ask me whether these techniques can replace programming jobs, I would say "no". The people/reporters tend to confuse ML techniques with automation and merge both since sounds nicer when they type the articles. The "automation" which is taking over is basically to write scripts and batch jobs to replace manual testing, sysadmin works in data centers and in certain other areas.

I was looking for this answer.

vivek_v wrote: What actually seems to cause the slowdown in Indian IT is the proliferation of FOSS software and tools with the requirement for engineers who have programmed and deployed systems using these tools with minimal resources, comfortable in multiple programming languages and have experience in cloud deployments rather than people management skills.

FOSS has always been there. Perhaps, the big clients of proprietary software are also now moving to FOSS. But you will still need programmers to code for the new set of tools. FOSS is not exactly free to build. Most of FOSS has for profit companies or consortiums behind them like RedHat is behind Hibernate, eclipse is paid for by a consortium of companies, so is apache. AFAIK, no big, game changing FOSS came out in the last few years. If anything a lot of new jobs were created by web 2.0, mobile and big data.

vivek_v wrote: I feel that the plight of organizations who supply expensive hardware with their software or vice-versa (leaving niche stuff like Teradata) would be even more terrible which is causing a chain reaction in the ODC's for those vendors run by Indian IT shops. No one nowadays wants an upfront huge Capex investment either in HW or in SW.

So basically, its the product companies that will be hit by FOSS. And IT (as in not programming) jobs because of ML.

I wonder if the Indian IT companies are not taking this opportunity to clean up their houses. There were talks of mas layoffs during the bank crisis too. And there are similarities in the number of people as a percentage of the work force are being laid off and the kind of people (mid level managers) being laid off almost 10 years ago and now.

I have noticed that the people most enthusiastic to become managers are technically weak people. They just want to have nothing to do with any kind of coding. The technically competent people even after becoming managers keep coding or at least involve themselves closely with design and implementation. In fact, the industry has moved away from pure managerial jobs more than a decade ago. Only in Indian outsourcing we had this pure managerial jobs (perhaps justified) and they seem to be feeling the heat.

kvraghav · Post by **kvraghav** » 12 May 2017 10:31

What type of jobs are we loosing in our organisation?
Database Admins, Network Admins because of cloud deployment.
Manual testers due to automation testing.
What type of jobs are gaining in our organisation?
Network security and devops due to cloud deployment,Big Data Analytics, automation engineers.
Programmers are stable.
Before we spell doomsday, one thing we have to remember, the new technologies need to be learnt whether by freshers or experienced guys. The only question is are they willing to pay that extra for your designing skills, domain and product knowledge. Even with machine learning and data analytics, we would like to have domain people to understand the data and program. So we preferable want experienced people to learn new technologies and get paid and sorry to say this, the current gen attitude puts me off. Call me an oldie if you want but i have seen that senior professionals work harder than freshers.

prahaar · Post by **prahaar** » 12 May 2017 19:28

I wonder, what prevents companies to "re-skill" a subset of their workforce with ML techniques. Based on my understanding, ML has shifted a significant chunk of earlier pure signal processing work into software development domain (this is an over simplification), especially if one is looking at utilizing currently available CNN DL libraries and building on top of it. There is significant application specific modeling and problem solving scope.

Marten · Post by **Marten** » 12 May 2017 23:17

We are seeing a massive push in new interfaces for decades old applications. Of course FOSS helps us save costs on large deployments, but for the regular deployments, it is but a marginal saving. The cost of rearchitecting the solutions will fund most of our units for the next 5-7 years! That's how much work is left in our space. There could be much more efficiencies to be gained, but quite a bit of the savings will probably be reinvested in improving the ability of the business to respond to changing customer needs.

Yayavar · Post by **Yayavar** » 13 May 2017 02:30

FOSS - someone still has to fund the engineers working on it. Big impact guys and many behind the scenes are employed by the Intels, IBMs, Huaweis, Microsofts even... it is an amortized cost across industry.

KJo · Post by **KJo** » 13 May 2017 03:01

So outside of all the gloom-e-doom, let's try this. If you had to think of ways to survive as an Indian IT firm in the current climate, what would you do? Maybe some new ideas will come up.

What would you do if you were Sikka or the TCS CEO? The old strategies will not work.

KJo · Post by **KJo** » 13 May 2017 03:09

kvraghav wrote:What type of jobs are we loosing in our organisation?
Database Admins, Network Admins because of cloud deployment.
Manual testers due to automation testing.
What type of jobs are gaining in our organisation?
Network security and devops due to cloud deployment,Big Data Analytics, automation engineers.
Programmers are stable.
Before we spell doomsday, one thing we have to remember, the new technologies need to be learnt whether by freshers or experienced guys. The only question is are they willing to pay that extra for your designing skills, domain and product knowledge. Even with machine learning and data analytics, we would like to have domain people to understand the data and program. So we preferable want experienced people to learn new technologies and get paid and sorry to say this, the current gen attitude puts me off. Call me an oldie if you want but i have seen that senior professionals work harder than freshers.

But won't you need people to administer the cloud deployment? Maybe the db and network admins can go there. I don't have any direct experience in this, but this is a possibility.
Large scale Automated testing is expensive and so only large companies and large banks can get into that. Mid sized and smaller companies cannot afford it as the investment does not make sense. So they would prefer manual testers who can also do automated testing like Selenium.

People think that soon machines will write programs and programs will be without jobs, but someone has to write those machines. Again a cost factor comes up for smaller companies and the technology is too far off to be deployed across the board. I am in the Data Analytics space and dabble in Machine Learning on the side and plan on adding that to my product in due course, but I don't see these technologies as a way to replace or kill off jobs en masse.

I agree with you, (maybe because I am an oldie too), senior professionals who started their careers before the Y2K era generally work harder. They have seen harder times when they were not in demand. The youngistan crowd in the last 10-15 years has only seen promotions and 20+% raises and multiple offers and have reduced their level of commitment to match this. This is only natural and a phenomenon that is seen everywhere, not just in India.

The next few years will be interesting.

KJo · Post by **KJo** » 13 May 2017 03:17

I joined a new company based in Toronto and went there to meet my team. It's growing rapidly and they are investing a lot into tech. The

I hear from the Engineering managers is that they need more people and cannot hire fast enough. The job market there is hot and so skilled people are in huge demand.
I wonder if Canada is different from the US because I have rarely seen this happen in the US except in the 2000 tech boom which I sadly did not exploit. This could be because of sudden demand for local candidates because outsourcing has reduced (?).

pandyan · Post by **pandyan** » 13 May 2017 04:19

With cloud, ratio of resources managed per head is substantially different. Cloud does all heavy lifting and you need people to configure and monitor. So one person can do the job of 10 or 15. Same thing with cloud developers

Neshant · Post by **Neshant** » 14 May 2017 04:01

KJo wrote:I joined a new company based in Toronto and went there to meet my team. It's growing rapidly and they are investing a lot into tech. The I hear from the Engineering managers is that they need more people and cannot hire fast enough. The job market there is hot and so skilled people are in huge demand.
I wonder if Canada is different from the US because I have rarely seen this happen in the US except in the 2000 tech boom which I sadly did not exploit. This could be because of sudden demand for local candidates because outsourcing has reduced (?).

The low value of the Canadian dollar vs the USD.

vera_k · Post by **vera_k** » 14 May 2017 07:45

ManSingh wrote:We do not have a competitive university system that responds to changes in landscape fast enough.
People might agree/disagree with the importance of this but I have seen universities with courses in 2017 like 8051 microcontroller and exchange server 2008. Compare this to latest courses from georgia tech or texas univs. They have self driving cars, ai, data science, am335x and what not.

Well, self driving cars is a 1980s technology that has caught on recently Alvin.

Its not so much about responding to changes, as a near total lack of research departments. Number of STEM PHds granted in India is quite low.

Neshant · Post by **Neshant** » 14 May 2017 07:54

Prasad wrote:While former ada and drdo heads decry the fact that components are imported easily and local players can't compete against a product made of tax free imported components.

If quick & duty free imports of electronic components is not implemented in India within the next 3 years, India will very badly lose out on the engineering front.

Making it hard to import components and tools that are vital for design and product development will destroy an entire generation of innovators.

Someone needs to get the ear of Modi-ji and let him know electronics park with rapid import & facilities is what India needs for takeoff in the consumer, scientific, defense & industrial electronics front. Without it, all the Make in India initiatives are a total waste of time.

With it, the vast talent pool of brainpower can begin to unleash its creativity and entrepreneuship.

The longer they wait, the further behind India falls.

NRao · Post by **NRao** » 14 May 2017 08:02

vera_k wrote:
ManSingh wrote:We do not have a competitive university system that responds to changes in landscape fast enough.
People might agree/disagree with the importance of this but I have seen universities with courses in 2017 like 8051 microcontroller and exchange server 2008. Compare this to latest courses from georgia tech or texas univs. They have self driving cars, ai, data science, am335x and what not.
Well, self driving cars is a 1980s technology that has caught on recently Alvin.

Its not so much about responding to changes, as a near total lack of research departments. Number of STEM PHds granted in India is quite low.

Nothing in "AI" is new, everything is pretty much from the 80s. The main diff is "speed" - faster CPUs, networks, larger-faster-cheaper memories, etc, etc, etc.

The foundation of "AI" cannot really change.

Singha · Post by **Singha** » 14 May 2017 10:22

Canada cashing in.

https://www.wired.com/2017/05/uber-hire ... ab-future/

My employer is reducing san jose and other locations in phases and giving work to ottawa.

Karthik S · Post by **Karthik S** » 14 May 2017 10:25

I heard Citi has moved many IT teams from US to Canada already.

vivek_v · Post by **vivek_v** » 14 May 2017 10:42

NRao wrote:
Nothing in "AI" is new, everything is pretty much from the 80s. The main diff is "speed" - faster CPUs, networks, larger-faster-cheaper memories, etc, etc, etc.

The foundation of "AI" cannot really change.

Three major things according to me personally caused the sudden explosion of AI,

a. As you had mentioned better computing power, more specifically GPU computing.

b. Large amounts of Data. Labeling it is a pain, but once you label using semi-automated or manual means, then things go much smoother.

c. Much better models of neurons (Relu/ELU..etc in place of sigmoids), networks (RCNN, RNN with LSTM), loss functions, training and normalization techniques (dropout, Batch normalization) ...etc and it's just getting started.

We don't really understand mathematically in terms of proofs why most what is mentioned in point (c) works except in terms of empirical ideas or hand waving reasons. This is a field of intense research to better understand these techniques from a rigorous mathematical proof point of view but this is not a reason to write off the same (or) be extremely antagonized towards the same. In areas where Deep learning could be used, it performs really well as long as the engineers working on the same are competent.

Anyways the field is growing rapidly and the number of really excellent papers in arXiv (very popular in data science community) is mind boggling and no time to read them all to try. If we leave our preconceived notions aside, this is the perfect time to go into Deep learning (a.k.a. Neural networks).

Singha · Post by **Singha** » 14 May 2017 11:14

Karthik S wrote:I heard Citi has moved many IT teams from US to Canada already.

Chaiwalla reports filtering on from massa of concerted efforts to reduce h1 in i.t. depts and hire local.

Deploy mav drones and see what you boys can dig up

Marten · Post by **Marten** » 14 May 2017 11:27

Our CO has clearly said no more H1s or L1s until the situation pans out. I had been given a final move order/option in Jan but knew there would be madness in a few months. Either ways, more jobs will move out of the US. Some to Canada, some to India.

CalvinH · Post by **CalvinH** » 14 May 2017 14:07

ManSingh wrote: We do not have a competitive university system that responds to changes in landscape fast enough.
People might agree/disagree with the importance of this but I have seen universities with courses in 2017 like 8051 microcontroller and exchange server 2008. Compare this to latest courses from georgia tech or texas univs. They have self driving cars, ai, data science, am335x and what not. Also GA tech has moved their masters in comp science program completely online and slashed cost to 7k usd besides increasing enrollment to thousands now. How the hell would we compete against such an offering.
It is time to wake up if someone is listening.

Apologies for the long rant.

agree about competitive university system but that data about university courses is not right. Even tier-3 private univs are into Java and Python in first year for comp sc/eng students. Its a different matter that they dont have good lecturers/professors.

In new tech young students in India are fairly exposed and aggressive as compared to when we were students, and working on Kbps internet connections.

NRao · Post by **NRao** » 14 May 2017 14:55

Much better models of neurons

How much of this is due to your first two reasons?

So, essentially, take out the first two and what would remain in the third?

In 1979-83, we simulated Chicago SMSA for 2005. The models took 27 hours on the then fastest machine: IBM 370 (no funds to use Cray). By 1983-86 I could run the same in about 10 on a UNIX based Altos desk top. Today, in real-time. GPU (self driving auto) itself make a 370 look like a kid without pampers. And custom chips, what can we say?

Anyways. Hope we use such techs for the right things. With all powerful nations being led by autocrats, never know what we will wake up to.

Can you please post a link to the site you mention about the papers? Thx.

vivek_v · Post by **vivek_v** » 14 May 2017 15:34

NRao wrote:

How much of this is due to your first two reasons?

So, essentially, take out the first two and what would remain in the third?

It's like three faces to a triangle if you remove two faces there won't be a traingle.

Having very little data would cause the models to get extremely overfitted unless you work in typical images where one can use transfer learning.

Having no computation power but lot of data would mean that you start the training on Monday and the iteration finishes on Friday.

Finally, without the discoveries in better ways of training the networks, half of the neurons would be dead (or) vanish (or) worse (i.e) never converge. I once tried explaining convergence to the management

.

Ideally, I would say that there is a fourth side which is the tools. Previously (say 3 years ago) mostly people used Caffe which used C++ or Torch based on LUA had a separate learning curve.

When Theano came up with python support is when people who worked in Machine learning (Sklean and rest) had an easy time in employing or at least try to experiment in Deep learning. Then again, finding something in Theano documentation was like trying to find a Chinese cab driver who can read your hotel's English address.

Then Google released the one framework to rule them all (i.e. Tensorflow) is when things became really easy since their documentation is top class. Nowadays when training a model, I spend 30% of my time in Keras (a pretty frontend for Tensor Flow) and 70% of my time in Tensorflow.

All I was trying to say that, something fundamental has changed and it is not the old hype cycle of overpromising and underdelivering in Deep learning. These things actually seem to work and somehow they model both linear and non-linear functions.

NRao wrote: Can you please post a link to the site you mention about the papers? Thx.

The website is arXiv in which the best and most recent articles would be published. For Data science this beats ieee or ACM hands down.

https://arxiv.org/

For a really sarcastic/funny take on all this learning material, please check out the below reditt thread.

https://www.reddit.com/r/MachineLearnin ... _learning/

NRao · Post by **NRao** » 14 May 2017 19:57

vivek_v,

The point being we actually did very, very well without the first two!!!!!! For a very long time.

And, I have not kept up, but doubt the ROI is great on most new findings. I suspect they amount to publishing papers to add to a resume. Am I right?

Thanks for those links. Will check them out ....................at leisure.

On trying to explain. In the 80s we had a very difficult time explaining a very simple statistical concept of "confidence". Our audience: 100% politicians. Ultimately we told them to let us know what they wanted.

Ever since this field of "Data Science" started (especially on the internet), the following captures my complain (from your reddit link):

Most people at Google are software engineers and don't perform analysis

There are head hunters that quiz you on Hadoop admin for a DS position!!!

pandyan · Post by **pandyan** » 14 May 2017 22:38

vivek_v - nice series of posts.

to add,
cloud based ml has made the "compute" accessible without heavy upfront investment and they also provide tools to easily analyze and assess quality of models.

second aspect is finding relevant features for complex non-linear fit. in lot of cases simple algorithms do work very well provided you have relevant features.

NNs etc where they shine is in areas where you need to try combination of higher order features (especially with 100s or 1000s of input features) that can be very difficult to try our own.

Singha · Post by **Singha** » 14 May 2017 22:40

What is the use case and domain you are discussing?

pandyan · Post by **pandyan** » 14 May 2017 22:47

Manufacturing, image analysis etc easily have 1000s of input features

Neshant · Post by **Neshant** » 15 May 2017 02:55

Apply for funding if you have a small company or if you are an academic or even individual with an idea for defence related product. It has to be something innovative worth funding.

---------

Defence Ministry to launch innovation organisation

NEW DELHI: Defence Innovation Organistaion (DIO) is to be launched as a non-profit company later this month to foster technology development and innovative products with commercial potential for the defence sector, informed sources said.

The company is being formed by defence electronics major Bharat Electronics Limited (BEL) and defence aviation major Hindustan Aeronautics Limited (HAL).

The Defence Ministry had in December approved the creation of a Defence Innovation Fund (DIF) under which the DIO is being formed.

"This was a Ministry of Defence initiative and was tasked to BEL and HAL. The new entity will work under the guidance of the ministry which will provide the ideas and topics for innovation," a BEL source told IANS.

The company will fund development of new and innovative products and solutions for India's defence requirements.

"Anyone from academia, medium and small enterprises, research and development institutes, individuals and start ups can approach DIO for funding," source said.

"BEL and HAL are contributing Rs 5 crore each to Defence Innovation Organisation, a Section 8 company, to promote innovation in defence. The company will be launched within this month," the source said.

As per the Section 8 of the Indian Companies Act, 2013, not-for-profit companies can be established for promotion of art, culture, science et al.

After the launch, DIO will choose a knowledge partner for screening the ideas received for funding support.

The selected ideas will be financially supported to work on the proof of concept and those will go through further down selection process to select concepts that can be funded for prototype development.

The successful prototypes will also be helped in their commercialisation.

"If we get good response then the fund can go up to Rs 100 crore, with Rs 50 crore each from both the partners," source added.

The process of the registration of the company is in final stages.

According to sources, the company's board will initially have two nominated directors, one each from the HAL and BEL. The process of selection of the nominee directors is underway.

The DIO, in future, may also have some independent directors.

http://defenceaviationpost.com/defence- ... anisation/

Singha · Post by **Singha** » 15 May 2017 10:18

do the big FOSS supporters like ibm, hp, oracle have teams working on large open src tools like cassandra, spark, storm, mongoDB etc who periodically upstream their work? they would be looking for people already having a good github profile not people looking to get into it.

that line of work and another line which is taking the right set of tools for a user requirement and glueing a pipeline of these tools (configuring, scripting, deploying) together and supporting that @ scale seems like sunrise sectors ?

so a section of web admins skilled in the latter must be doing really well.

SRoy · Post by **SRoy** » 15 May 2017 14:50

With FOSS tools one has to be an administrator and a developer rolled into one.
On *NIXs the dependency hell is way lot worse than Windows DLL hell. Unless there is something wrong with basic OS packaging itself, dependency issues can be resolved by building components from sources. An average "admin" cannot always do that. A developer can do this in minutes.
Then of course the possibility of customizing a FOSS software which a developer alone can undertake.

There are such admin positions however which are a waste of time and money. We used to have a JIRA admin. He handed over the controls and moved to something worthy the moment we created the first project in JIRA.

From a maintenance POV FOSS tools are very scriptable are do not require baby sitting. I still have occasional updates from ex-colleagues about few systems running without a glitch. Huge code bases in GIT and TFS, pulled out by a CRON job, revision header auto generated, output built and run against test scripts and culprits ( well joking ... ) identified from commit logs, defect reports mailed overnight to all concerned. All from a set of CRON jobs. Long time ago, much before the arrival of DevOps hype. And I think the effects were visible to us, because that infra setup lead to certain people redeployed and vendors offloaded. Clearcase was discontinued along with the admin postion that was tied to it.

Further on, there are product companies that have their offerings itself built around customized FOSS, so a dedicated admin won't help.For the enterprise comm business we used to maintain a customized Asterisk version for many cost sensitive customers segments.

I believe more than the companies, individual techies should outright refuse or avoid going into these admin type roles. Doing so would trap them in low skill pits as have already happened with COTS experts.

Do companies scout for talent in GitHub? They do in StackOverFlow for sure. You need to have an expert developer in the recruitment team to crawl though GitHub content. But, lately I have been noticing a job alert section in my GitHub dashboard. For us techies these two platform are way far valuable and important than a LinkedIn profile.

Singha · Post by **Singha** » 15 May 2017 16:34

how are people able to upload stuff into github? all of what code i write is proprietary and same would be case for many. do people upload personal projects there or get into some bigger project by picking up lists of fixes and enhancements which the owners put up?

SRoy · Post by **SRoy** » 15 May 2017 18:25

@GitHub,

I have few personal projects and one FOSS project where I used to contribute in spare time. For the FOSS contributions, I had to obtain permission from the legal. FOSS part made sense to the legal because, we used to maintain a customized fork of that stuff. Have stopped working on that, but that project still appears in my dashboard (good showcasing).

Personal projects are a different case altogether.

I have classes/libraries/entire frameworks (all sorts of network protocols, AV codecs, GSM specific parsers) developed over a decade and more, fine tuned, simply too valuable to be thrown away or forgotten, they have since surpassed NDA clauses where they had origins in corporate tech. So, wrapped them into little showcase applications with nice GUIs. The GitHub repository also acts as an insurance against hard disk crashes. This stuff is a backup of my important work and in future would serve as my personal portfolio.

Check out for old work that is of interest and whether it still has active confidentiality agreement or not. If the work is of interest and of no use to corporate, then take it over, nurture and publish.

Second type of personal projects are exploratory type. Pick a new tech/domain, think of an use case and when have time, write a small application. Everyone writes these kind of applications, trying some new libraries, some new domain, sort of 'Hello World" in area XYZ. So, why not take a copy, clean it up, extend and make it useful.

Therefore, I believe it does not take a lot of time to build a personal portfolio.

The way IT is moving into, I guess we have to transform into lean and mean SF operator type, maybe even mercenaries. Days of lazing in trenches waiting for the next salvo are over.

Yayavar · Post by **Yayavar** » 17 May 2017 20:39

There is BE in Software Engineering being offered now. I could not tell if it is an overkill for 4 years. It shares the base CS courses but then diverges to design methodology, testing, project management etc. Or is it better to still get a core engg (assuming CS is not an option) like EE/EI or Mech with programming skills on the side (assuming all prob have some programming and computerization included). Am being asked to advice the upcoming crop and find myself a little short on info.

Singha · Post by **Singha** » 17 May 2017 21:05

I would say core engg...avoid soft fluffy electives meant for future pptgiri. One need not sell oneself to the dark side at ug level imo.

Mech chem aero ee ece is a mass of tough math . Great..embrace pain now ... now is the time for young hoplites to answer the call hu hu hu than sit in a tent 10miles behind thermopylae being a budding product mgr

Those willing to embrace death shall live
Those afraid of death shall die...

Yayavar · Post by **Yayavar** » 17 May 2017 21:18

kya baat hai Singha saar... It makes a lot of sense.