Managing a team of AI experts is a challenging yet rewarding endeavor, and Vivek Gupta, an AI team manager, shares his insights on cultivating strong machine learning engineers. In his talk, Gupta emphasizes the importance of feedback, learning time, and cross-team collaboration for the growth of these engineers. But here's where it gets controversial: he suggests that engineers need guidance not only on technical skills but also on interpersonal interactions.
Gupta believes that engineers, fresh out of school, crave feedback to improve. This feedback encompasses their coding abilities and, surprisingly, their social skills. It's a delicate balance, as engineers must learn to collaborate effectively with others and understand the value of diverse perspectives.
To foster this growth, Gupta encourages managers to provide learning opportunities and mentorship. Senior engineers can play a crucial role in guiding their junior counterparts, and by coaching these mentors, organizations can scale their knowledge transfer effectively.
But it's not just about coding; engineers working with AI and machine learning in production environments must also grasp data management. Gupta highlights the need to track and manage data used for training and validation, ensuring consistency and efficiency.
And this is the part most people miss: the importance of human-in-the-loop validation. User feedback is key to closing the loop and improving model performance. It's not just about giving a thumbs up or down; it's about providing constructive criticism to enhance the model's capabilities.
In an interview with InfoQ, Gupta delves further into his team's practices. They host regular hackathons and dedicate learning days after each sprint. These sessions focus on both technical skills and career management, offering a well-rounded development experience.
For senior engineers, collaboration involves cross-team knowledge sharing and leading learning sessions for newcomers. It's a delicate balance between technical expertise and mentorship, and Gupta believes this approach fosters a natural technical leadership structure within the team.
When it comes to managing large language models, Gupta acknowledges the challenges but emphasizes that the principles of MLOps still apply. Fine-tuning these models requires careful data management and prompt evaluation, ensuring a well-engineered approach for production scenarios.
So, what do you think? Is providing feedback on interpersonal skills as crucial as technical guidance? How can organizations strike the right balance between technical expertise and soft skills development? We'd love to hear your thoughts in the comments!
