Dealing with Engineer’s Pain Points

There is no day in the life of an engineer without resolving problems. Often, we, as engineers, focus on building new tools or delivering increased performance of an existing function (usually ten-folds). This focus on the new and cool is great in the fast paced areas of emerging business lines.

However, once we develop a business function or a capability, we need to quickly operationalize it and optimize it to function well. When things go wrong with that operational business function, we need to quickly identify the underlying problems and correct them to avoid wasting time and money that adds no value to the platform users.

In the end, if we can’t deliver on our basic processes, no one will trust us with anything else. This is where problem management is crucially important. Here are the 5 basic steps of problem management:

1. Problem Identification
2. Categorization and Prioritization
3. Root Cause Analysis
4. Resolution
5. Closure

The first step in solving problems is realizing you have a problem. Problem consciousness is really important. In the early stages of problem management, problems are everywhere. There is typically little effort required to find something broken and attempt to fix it. Some organizations use an incident management function to feed their problem management process. They review and postmortem every major incident to find problems, expose their root cause, and drive organizational change to fix them. Other problems are harder to find when the systems and volumes of incidents become more complex.

Some operational entities for instance, perform about 400,000 server repairs a year. These repairs come in the form of server repair tickets. Often, the tickets are driven by various factors like component quality, software tooling, software automation, or even transactional defects. It is challenging to spot the systemic problems when they are buried in a sea of data.

Once a problem is detected, one needs to quickly categorize it to help understand its significance and the appropriate urgency required to resolve it.

A few basic categorizations of a problem include:

• Impact of problem (pain point): How does this problem prevent us from operating normally?
• Value if resolved: When presented with a problem, we can either ignore it or attempt to fix it. Some problems are so small, we ignore them. They are not worth fixing. Other problems are huge and need a solution right damn now! Thus, it’s important to weigh the value of the problem.
• Problem complexity: When we try to prioritize which problems to fix first, it’s really helpful to understand the complexity of the problem at hand. Sometimes complex problems require a great deal of support and preparation to solve. Others, not so much. It is good to become a big fan of solving those high impact, low complexity problems first.
• Availability of required skill sets: Solving problems in a big complex organization requires people who are knowledgeable in the key problem areas. It is super helpful to know, at a high level, what resources are required to solve the problem. Possibly keep an inventory of those resources to help avoid wasting time searching for them later.
• Estimated time and cost to resolve: Often a rough order of magnitude (ROM) is all that’s needed during the categorization phase of managing a problem. Form a combat math estimate of the time and resources required to resolve the issue.

Once you have the problem adequately categorized, move on to the next phase.

1. Paper repro: “Learning to Learn by Gradient Descent by Gradient Descent”

2. Reinforcement Learning for Autonomous Vehicle Route Optimisation

3. Basics of Neural Network

4. AI, Machine Learning, & Deep Learning Explained in 5 Minutes