Building Machine Learning Tools at Cohere.ai

Michelle Watson

Machine learning is an exciting field with new startups cropping up to make it accessible for businesses. At Techintern.io, we provide a platform for exciting new companies to scout promising and talented prospects in the industry!

Michelle, a Biomedical Engineering senior in University of Waterloo, was successfully hired as an intern to work on machine learning at Cohere through Techintern’s platform. Cohere is a startup that aims to build machines that can effectively understand language by taking into account the syntax, semantics, and even context of the words inputted in the system.

Internship Experiences

  • What did you work on over the course of your internship?

Michelle worked mostly on external documentation on the website. Despite being a junior of the team, she had full ownership of creating a data pipeline for a specific data type. Aside from that, she was also tasked with providing support in the general development and testing of future pipelines that can be streamlined.

She also worked on the fine-tuning pipeline by making it more robust and further optimized to work better with their new training queue system.

  • What was unique about the company and role?

Michelle found it unique considering that the company is still young and therefore open to new ideas. The role was unique for her as well that she could branch out to other areas such as data engineering despite her primary role in infrastructure.

She worked with the product team, who was also relatively new due to Cohere’s stealth nature. What interested her the most with the unique role was the process of productionizing machine learning since her related experience prior to internship was more on the academic side.

The team was also very social, having a weekly chess tournament and coffee chats of about 2-3 people randomly meeting for 15 minutes daily. Meetups were also conducted that further contributed to a more open social culture.

Michelle particularly enjoyed the onboarding process where she would go on meetings with new members so that it was easier and comfortable for her to transition into engaging with the rest of the team members. Cohere also provided paid lunches through Ritual.

  • Do you have any tips/advice for tech interviews? What was your interview process like?

Michelle elaborated that the interview process had 2 rounds. The first round was more behavioral and laid back, and she could ask general questions about the company since it was still in stealth mode during that time.

The second round, she said, was more technical as she was required to show them a project she worked on prior to internship and showed them how she would solve a coding problem through a Google Doc.

Tips/advice

For students who are preparing to take on their internships soon, she advised you ask about the interview process in advance so you can prepare smarter. She added that it's also good to look up your interviewer online to see what you have in common so you can establish a good conversation.

When asking questions, she said that generic questions are good, like team size and their culture, as well as the stack and tools that they use daily, but you also have to ask them something specific that you can't look up on the internet. She usually asked about internal documentation so she knew how to proceed when on the job based on the work by others in the past, so she advised doing that as well.

She also asked about compensation when she was in the interview, but she clarified that it's at your discretion whether you want to ask it or not. For tech heavy roles like what Michelle did in her internship, she advised getting as much information as you can about the specifics of the role.

She encouraged asking a lot of company-specific questions because you also have to know if you're a good fit for the company. If the company is still in stealth mode, like Cohere was when Michelle applied to their team, you can ask about companies similar to them.

  • What did you hope to gain from your internship? What did you gain?

Michelle said that she hoped to learn more about productionizing machine learning. Although the framework was already there mostly, she was able to gain a deeper understanding of Go methodology, Kubernetes, Docker, and Redis which could all be used and work together to create a data pipeline.

  • Do you have tips for future interns?

Michelle emphasized not to be afraid to ask questions. She said that, although it’s better to ask the questions early into the internship, you shouldn’t be worried about asking questions later on that you may think “I should have known this by now.” As long as you get the information you need to do your job and be productive, it’s better for everyone to just ask.

She also added that you have to solve your own problem before asking about them. Even then, there’s no harm shooting a question out there and continuing to work on it as you wait for their response. If you figure it out on your own, great, you just have to send them a follow-up message saying that you figured it out. If not, you can elaborate more on your message and provide details of your attempts to solve.

Conclusion

  • Michelle is currently in her senior year in Biomedical Engineering at University of Waterloo. We helped her connect with Cohere, a Natural Language Processing (NLP) toolkit that aims to help technology advance so that it’s easier for computers to parse the syntax, semantics, and context of the inputted words.
  • She worked mainly on the external documentation of the website and infrastructure of data pipelines and they gave her full ownership of a specific data type. She could also branch out to other areas of the toolkit’s development, like data engineering.
  • Michelle prioritized learning by having the courage to ask questions when she ran into problems or if there was something she didn't know yet about the job.
  • To better prepare yourself in the interview, try to figure out the exact interview process in advance and research on the background of the interviewer so you can develop a strategy of how to present yourself. Ask both generic and specific questions about the company.

If you're interested in hiring tech talent like Michelle, we'd love to have you join us at Techintern.io! Sign up at https://www.techintern.io or feel free to reach out to kingsong@techintern.io if you have any questions.

If you’re a student looking for your next internship, check out Techintern.io! Our common application process is designed to make applying as easy as possible, so you can find great internships like Michelle did. Sign up now at https://www.techintern.io/student/signup.