DSA by domain
AI & Machine Learning
The roles building models, vision and language systems — where the algorithmic core is math-, matrix- and DP-heavy.
What the roles here have in common
2 topics are core for every one of the 4 roles in AI & Machine Learning. Whatever job you end up interviewing for inside this domain, this is the part that was never optional:
11 more topics are core for some of these roles but not all of them. Which ones apply is exactly what changes when you pick your role below:
Roles in this domain
Same 370 problems, re-sequenced per role. Pick the exact one you are interviewing for. Every role opens its own roadmap, and the roles with a full guide open the guide.
- MLML loops tilt toward math and DP-flavoured optimisation rather than graph traversal: 6 core topics, 125 of our 370 problems, with matrices and math sitting in the core tier. Expect the DSA round to be mixed with applied stats and modelling questions elsewhere in the loop.125 of 370 problems sit in its core topics, starting with Arrays, Matrix, Math.
- Computer Vision EngineerImages are matrices, so linear algebra, kernel windows, flood-fill graph traversal, and stereo/seam DP dominate; string and trie topics are near-zero.165 of 370 problems sit in its core topics, starting with Matrix, Math, Arrays.
- NLP EngineerTokenizer tries, Viterbi/edit-distance DP, beam-search heaps, and parse-tree recursion: string-heavy in exactly the way computer vision is not.147 of 370 problems sit in its core topics, starting with Strings, Trie, Dynamic Programming.
- Robotics EngineerMotion planning is graph search plus kinematics math and rotation matrices, with priority queues driving A*/Dijkstra; string and trie work is absent.131 of 370 problems sit in its core topics, starting with Graph, Math, Matrix.