Gather requirements. Find out exactly what the team is looking for. Spend an hour reading up on the topic.
Look to see if there's an open source solution to the problem with a compatible license.
If not, figure out the simplest system that solves part of the problem. For exception handling, this could be a single try-catch around your entire program that logs the exception and restarts the program. Congratulations, you've now avoided failing the task! In my experience, it's much better to present a simplistic solution that barely addresses the problem then a well designed solution that doesn't work.
Now that you've made sure you won't get to the deadline with nothing, iterate. Make the logging configurable. Add some inner exception handling for cases that can be gracefully recovered from. Read more on the topic to see what other people have built.
Again, try to get each iteration working as quickly as possible. When you're in territory you've never dealt with before, you want to work in short sprints. A few days of coding and testing, and the system runs and is slightly better than before. Repeat.
