The Signals in the Noise
How 5 women are blazing the trail on Salesforce’s Data Intelligence Team
Women currently only make up 26% of data science professionals. However, this summer at Salesforce, I was lucky enough to work on the Data Science Apps team (a subset of the Data Intelligence team), where 50% of my team members were women. I had the opportunity to learn from 5 talented, passionate women* who have established unique roles in this industry and continue to stand out. This post is aimed towards new, aspiring data scientists who want to know more about how to be successful in this field.
Lesson # 1: Find your edge!
M.I is passionate about models and discovery. Coming from an academic background in Physics, M.I explains that the transition to this field felt organic as she already loved model building and making insightful recommendations. Whether you’re in HS, undergrad, PhD or have other industry experience, it is always a right time to explore data science and pick up the skillsets you need.
She attributes a lot of her professional development to relationships with mentors and teachers who have provided her invaluable feedback as she aided them with their research. “Make sure your relationship is symbiotic and you give as well as take to truly allow it to flourish.” Take advantage of research/work opportunities in school and university as those experiences allow you to get in the habit of sharing all results and drilling into the strengths and weaknesses of your models’ assumptions.
M.I successfully demonstrates this trait in her presentation on monthly time series modeling and forecasting where she was transparent about the model’s limitations and the new edge cases she discovered in the process. This communication strengthened her presentation by allowing team members to avoid these explicit cases in the future, a valuable contribution for any data science team.
Lesson #2: Iterate ^ n
In elementary school, I.Z knew she liked numbers and coding. Since then, I.Z has iterated a lot in her own career to get to her current position. Starting off as an Applied Math and Economics major, she soon realized her passion lay in engineering and not so much in finance. Post that, she has worked as an analyst, a visualization expert and is now a machine learning engineer.
She has applied the belief that iteration is the key to the products she has created. While developing Salesforce’s Einstein Guidance (EG), her team went through numerous modulations to ensure that the product is extremely accessible and usable by all its stakeholders. “Even after a product is built and pushed to production, the changes don’t stop.” She spent a lot of time redesigning various features ranging from color schemes and positioning to discussing what metrics should be prioritized and which should be discarded. This insight comes from conducting user research and being extremely open to receiving constant feedback and implementing the changes quickly.
Applying this to her personal skillset, I.Z explains that she is “constantly exploring new tools and algorithms.” Becoming comfortable with new programming languages becomes easier once you abandon your mental blocks and get your hands dirty with programming and mathematics and constantly learn from your mistakes.
Lesson #3: The Power of Experimentation
X.J traded in the world of theoretical math for applied statistics. Beginning her career at a time when the title “data scientist” didn’t exist, X.J experimented with different roles such as business modeling analyst before getting to Salesforce. “An analyst performs a lot of descriptive statistical work whereas a data scientist focuses on statistical modeling to predict and make recommendations.”
She explains that her team remains successful by performing plenty of experiments. “Not all of them work, many fail, but your teammates benefit most from this transparency and communication so that they avoid going down the same path themselves.” The bigger goal is to create a usable product for your end customer (X.J led efforts on Salesforce’s CloseIQ, a major component of EG). Simplicity and interpretability are two values that X.J heavily emphasizes on. On a data science product team, no matter how complex and game-changing your experiments are, if you can’t convincingly communicate your findings to your customer, the product you are creating loses a majority of it’s value.
X.J is extremely excited about new developments happening in NLP and hopes to, one day, experiment enough to build her own accessible product that will improve her users’ daily life. At the same time, she remains a curious learner and advises that readers take advantage of the vast open-source material that is available to keep updating their data science skillset.
Lesson #4: Respect yourself and your code
As a child, lead machine learning engineer C.G wanted to become a Nobel-laureate. Even though she hasn’t received one yet, C.G deserves respect not only for her expansive technical skillset and ability to create robust and stable architecture for data pipelines but more-so for her diligence and passion to create readable, structured, well-documented code.
At work, she strongly endorses exhaustive code review processes between her teammates and herself. “My team and I review everything from variable name changes to suggesting complete data structure modifications.” A software engineer at heart, C.G stresses the importance of being organized and clearly documenting your code at every step. Plenty of editors such as IntelliJ can give you errors if your code does not fit into a specific coding style so as a new programmer, become comfortable with using such tools to make programming an art.
After exploring different worlds ranging from context aware computing, anomaly detection and forecasting, C.G hopes to now find a specialization in one domain while also continuing to keep up with the rapid progress in the rest of the industry. If anyone can manage all of that, it’s probably going to be her.
Lesson #5: Take the lead, don’t wait to be given it
A.N escaped the war in Ukraine to move to San Francisco, bringing with her only her knowledge of computers and math. In a short time, she was able to establish herself as a senior data scientist and make a new life for herself in the United States.
Today, she operates as the official scrum master on the DSA team helping facilitate the team’s operations, alignment and solve blockers that members face. In addition to participating in modeling efforts towards the Einstein Guidance platform, A.N also takes initiative to help other teams tackle their big data science issues. She says “to be a leader in a data science team, you have to be proactive and find problems to solve yourself. You also have to be able to sacrifice some of your priorities for the larger team’s but it will pay off in the long run.”
A.N was able to be succesful in the data science world after having lost pretty much everything. She secured her place through building valuable connections and mentors in the industry and by constantly reading and absorbing articles and blog posts. She embraces the automation era and says it “will give her more time to experiment with more fascinating concepts while allowing the machines to do the tedious work.”
3 Common Takeaways:
- As a newbie, start off as a full-stack data scientist. Explore all aspects of engineering, modeling, visualization, and analysis before choosing where to specialize. This will make you invaluable if you choose to join a smaller company where there are limited resources and you can grow quickly to lead all data science efforts.
- As women, use positive stereotypes to your advantage! Women can be extremely creative, pay attention to detail, and can multitask which are essential skills of any role in the data science world.
- Start early, learn quickly. As soon as you decide this field excites you, be open to fostering a technical mind and catching up on the fundamentals in math, statistics, programming, and business strategy. Use resources such as Coursera, Khan Academy, Kaggle, and Codecademy to help you on the way. Apply for internships and even if it’s not a data science specific role, make sure to incorporate as many of the skills you’ve learnt in any job you choose to pursue.
I hope this article was helpful! If you have any further questions for me or any of the professionals featured here, please shoot me an email at schaitra@u.northwestern.edu.
*First and last names of these individuals are not shared to respect their privacy.