Forum Replies Created
-
Posts
-
It’s a small world….I work in Murry Hill.
I will bring up about using one of the AI models to optimize the code base. Not sure they will buy it. But this might be a great task for a co-op or new hire. Yes take a part of the code base and see if the complexity (which I assume will be measured in lines of code) can be reduced. This might be a real eye opener. Cool.
My job is transmission system test for Optical systems that are used all over the world.
The company has two main software releases per year, in addition to point releases to address field issues.
My role is to verify Fault and Performance Monitoring (PM). Each release I am given features to test. Usually the feature is a new Line Amplifier or Add/Drop module. Or an enhancement to an existing pack which would entail adding additional channel widths or combining channels.
I verify that the PM counters and telemetry for the new packs are reporting as expected and the alarms are reporting as expected.
I review the requirements for the new feature. Then create or modify test cases for the pack(s). When the new packs are received I usually test them manually using a command line interface or Yang model commands. When I verify the packs are working reasonably well I start the process of automating the tests. Using the commands I run manually I work with the automation team in another location to get the scripts working on regression test systems. I basically architect the tests and automation team writes the test and gets to into the companies APT automation system (I am not a great coder so I depend on my automation person). Some tests are new but most are variations on existing tests. Once that is complete and we verify it works well the tests move to regression phase for the next release. The team in India runs regression. As the tests are run they call myself or the automation team if there are problems.
I am the old guy who likes to go to the office everyday. Not many do that any more. But I like to be by the hardware and show the young engineers the job. Funny that young people come to the office more than old people!
The code base for the systems is all C++ custom that has been used and reused for twenty years. So you always have the people who know how the spaghetti code works and the new ones trying to figure it out. System engineering in many cases just document what the developers have written. So getting on well with the developers in the know is a must.
Systems like Splunk are used to track test regression tests and flagging tests that fail. Failures are mostly script issues or a problem with the configuration or test equipment.
I guess I could go on and on. I can see the whole program could use a revamp. But you know the story Hardware comes in late. Code is late. But the date must be met so you make it by reducing the test cycle. So everyone is rushing and there is little time to think of improvements. I’ll stop here.
That all said I really like my job.
I’m trying to think of a way to apply AI and prompting to my current job.
I test software on optical systems. The prompts we talk about are really interesting, but to not seem to apply to what I do.
Maybe when the code base and requirements are uploaded (that won’t be anytime soon), the code can be simplified and enhanced and we can develop new simulators to test.
It’s just hard to think of a way to apply AI besides putting in a python script right now and having gpt check it.
Maybe I need to open my mind a bit more. Looking forward to Prompt University.
