As SWE in a Big tech, which DSA do you apply mostly in your day to day?

Trees. I was working with some guide data (in the form of decision trees) from a vendor where the API response came as a raw list of nodes with IDs attached to them.

The goal was to do a LLM-based enrichment scheme by feeding in the leaf nodes to a language model and having it extract key facts. In this case, getting pathing information on the leaf node was extremely valuable. (think URLs like abc123.com/products/trucks/tires)

So I found a way to identify the root node and did a BFS traversal to save the "filepath" of leaf nodes. This made the LLM-based analysis much better. Didn't think much of it until the vendor asked a few months later about how I did it 😅

Haven't worked in big tech yet but yeah.

Other quick examples conceptually:

• you can build a graph at scale using Redis. Since graphs can be expressed as sparse matrices, and sparse matrices are implemented on computers using hashmaps, and Redis is a gigantic hashmap/KV store

• Never use set() in Python, you get nondeterministic results. Instead use the OrderedSet object

• Workflows are generally DAGs, ie Airflow or even LangGraph. The better you construct workflows as composable DAGs, the smoother these are to debug/extend