Demonstrate clear thought process & reasoning when suggesting an approach.
Develop clean & elegant solutions to technical problems. Avoid force-fitting of same model/approach for every problem & over-complication with exotic-sounding approaches/models.
Find, read, digest & explain related literature for specific problems (arxiv etc.) & suggest the best direction to take
Ownership of model development (Computer Vision algorithms & Deep learning models)
Ownership of model improvement with scientific & logical analysis of model failures
Develop clean code for reproducible experiments, use git/dvc for code/data/model version control & MLflow to trace the results
Present, in detail, the model performances and limits to stakeholders
Fully understand the frameworks/models being proposed/used
Go deep into project to identify & frame clear problem statements/ open challenges in current work
Identify & point out risk & challenges, suggest mitigation and scope the work accordingly
In-depth written analysis of approach / problems / challenges / solutions
Technical review of other team members\' work
Animate technical discussions in various forums
Identify re-usable components/framework/assets
Help evaluate vendor products /solutions
Help with interviewing candidates for further growth of team
Technical tutorship and/or mentoring of less experienced people