Adhithyan Vijayakumar

Hi. I am. I was working as a back end engineer full stack back end engineer for about 6 years, uh, in startups, big companies, etcetera. I have worked in wide variety of software development life cycles ranging from planning to development, to maintenance, to testing, to deployment, and debugging issues in production. I have also Did front end development, so I can do both back end and front end. But throughout my career, I have worked on full stack roles, but it was a back end heavy role.

I think some of the libraries that are used in finance are Pandas, NumPy, Saipay. Uh, so, for example, Pandas help in CSP manipulation. Uh, there is an inbuilt library in Python, uh, that is called a CSP. Uh, like, the inbuilt library in Python takes around 5 lines to open a CSV file, but with pandas, We can just read a CSP file with a single line like this. There are many such useful operations in pandas and NumPy that helps to simplify while we work with large datasets.

Uh, as it refers to, uh, consistency, isolation, durability. So atomicity refers to, uh, each transaction is independent from other transaction. Either a transaction happens or it does not happen. For example, while during a transaction when power goes up, uh, the transaction state won't be recorded in the system. That is atomicity. Consistency is, uh, when, uh, when we When we enter data, it stays in the database. As long as we don't modify the data, it stays the same. So What we are entering into the database is the data that we receive when we, uh, issue a query. Isolation, Each transaction should not affect any other transaction, so every transaction should, uh, execute within uh, if within a sandbox. So, uh, each transaction does not know other transactions are occurring in the system. So durability, uh, whatever data we input on the system, it stays there. The data is not corrupted. It stays there as long as we delete in the system.

Self, uh, cloud computing, continuous integration helps to maintain the quality of code, ensure that new features does not break the existing features like, uh, we'll be writing unit test, integration test, functional test. So continuous integration pipeline will check for the quality of code, uh, file find and tell what lending issues are there. Run our test suite so that, uh, we'll get the test coverage and ensure that, uh, the code is of high quality and nothing is breaking. Uh, in terms of continuous deployment, many companies are opting for 2 week 2 week development life cycle. So they plan the start of 1st week, 1st day of 1st week, and they develop the things in the 1st week. Then the 2nd week, they give it for testing. At the end of the 2nd week, they'll go they will deploy it in the production. Discontinuous deployment ensures that new features, new bug fixes are getting released constantly, then the system is getting improved over time.

Uh, first, uh, I I understand that in finance finance the finance product, um, getting real data or logging the user data will be will be causing privacy issues. So logging won't be an option here. We'll be having, uh, uh, the system monitoring systems like New Relic and, uh, some error tracking systems like Sentry. We can get the error trace and try to, uh, debug why the error is happening, whether it's due to an run time error, or whether it's due, uh, whether we have, uh, failed to handle case, edge case, or or if it's due to an external library that we are using are, uh, it's due to an irrelevant data or something like that. 1st, we need to identify whether the issue is being caused by our code or third party code. Once we find, uh, once we found that it's a third party code, we can raise an issue in there, but but tracking system, and we can fast track the fixing that issue. If it's an issue from our side, we we can try to fix it as fast as possible and uh

I have used Lambda to enhance performance and responsiveness of a Python application. Uh, Yeah. I have deployed the APIs in a serverless land application because, Uh, APIs won't need a constant running server and Lambdas being serverless. Uh, it helps, uh, to give the response in a very, very quick time. So, Yeah. I have used the AWS Lambda to enhance performance and responsiveness of a Python application.

So the given code is a recursive function. Uh, so it has 2 blocks. 1 is terminating condition for the recursive block that returns 1 if the input number is less than or equal to 2, and the else part is the recursive block. It branches into 2 division, n minus 1 and n minus 2. I think the function, uh, function tries to calculate the the function, uh, will return the value of n when n is given. For example, we are passing calculate 3. So, uh, it goes to if block 3 is not less than 2, so it goes to a else block. So here, the recurs and branches into 2. Calculate 3 minus 12. Calculate 2 plus calculate 3 minus 2 calculate 1. So 2 branches are called calculate 2 calculate 1. So here, calculate 2 is call. So, uh, we'll go ahead in the recursion again. N is less than or equal to 2. It will return 1. So we are going here again. We we already have 1 records and block calculate 1. N is less than or equal to 2. 1 plus 1 returns to. Uh, okay. Then, uh, let's, uh, add another example. Calculate 4. It goes here. Calculate 3 plus calculate 2. Calculate 3 plus calculate 2. 2 plus 1. So this function returns the value of n minus 1. When we give 3 as an input, the answer will be 2. When we give 4 as an input, the answer will be 3.

Yes. Uh, in the try block, we are returning, You're returning a JSON that is a dict object, but in the accept block, we are returning an exception. So, uh, the return signature is not matching. Either we should return a dict in both cases or string in both cases. So what we should do is In the Excel block, we should change the return e to, uh, take, for example, return, open parenthesis, key message, colon key. So we should, uh, give like that. Uh, and, also, I am thinking of 1 improvement. Before return response to JSON, we should check if the Response has yielded a 2 x x's response. Whether the response was successful and yielded a status code between a 200 to 2 not 4, You should check them using if response dot okay, return respond dot JSON. So we should do that. Uh, it will also improve the code.

Uh, so in in the context of cloud architecture, Comp components should be, like, uh, should be loosely coupled. 1 cabinet should not be tightly coupled with each other. So, uh, in one of my companies, I was working on this big Python project where everything was tightly coupled. So Using this architecture, uh, I kind of suggested to the management to, uh, disintegrate everything, uh

So, uh, the one way I am thinking of is that, uh, while integrating data from various sources into an unified system after inserting data, we should run a background job to check if the data inserter, uh, is being present. Uh, is The data inserter and the data that came from the data source is the same. So we should, uh, verify this once by running a background job. And, also, we should ensure that