The 125th episode of Datacast is my conversation with Sakib Dadi - a Vice President at Bessemer Venture Partners. He focuses primarily on early-stage investments in developer platforms, data products, and software infrastructure.
Our wide-ranging conversation touches on his undergraduate experience at UPenn; his decision to pursue venture capital at Bessemer; his investments in developer platforms and data products; Bessemer’s roadmap to data infrastructure; trends in the data engineering, data science, and metadata management ecosystem; the evolution of ML infrastructure; his process to shape investment theses and build empathy with founders; and much more.
Please enjoy my conversation with Sakib!
Listen to the show on (1) Spotify, (2) Google, (3) Deezer, (4) RadioPublic, and (5) iHeartRadio
Key Takeaways
Here are the highlights from my conversation with Sakib:
On His Upbringing and Academic Experience at UPenn
My upbringing was heavily influenced by my parents, who immigrated from Pakistan to America around 29 or 30 years ago. They eventually settled in Southern California's Orange County area. Growing up, they instilled in me a strong work ethic and encouraged me to pursue various interests in school, including the sciences, math, history, and literature. They wanted me to have a broad and multivariate education.
As I progressed through elementary and high school and began applying to colleges, I knew I wanted an education that crossed multiple disciplines. My cousin went to Penn for his MBA and told me about the M&T program, a dual degree in engineering and business. Given my interests, he thought it might be a good fit for me. After researching the program, I applied and was fortunate to be accepted.
The M&T program provided the perfect fit for me. I was exposed to the practical applications of an engineering degree with a focus on technology and its broad applications in business and other sectors. The program tied the business aspects of technology with the sciences and math, giving me a well-rounded education.
On Becoming Interested in Startups
When I first started exploring my interests, I wasn't sure what I wanted to do. I knew I was interested in science, math, and other subjects, but I didn't have a clear direction in mind. At one point, I even considered becoming a doctor.
As I began my studies at Penn, I started to explore different areas and eventually found myself drawn to the startup world. This was around 2011 to 2015 when the tech industry was rapidly growing but not yet as well-known as it is today. Companies like Facebook and Google were already making waves, and people were excited about the possibilities.
For me, it was the early signs of growth and change that really caught my attention. The startup world was evolving quickly, and I wanted to be a part of it.
On His Favorite Classes at Penn
I really enjoyed my time taking classes in both engineering and business, as well as pursuing my passion for history. One of my favorite classes was a Roman history course, which fascinated me. I loved looking back at world history and piecing together the events that occurred, such as Julius Caesar's assassination and Octavian becoming emperor. The professor did an amazing job of providing primary and secondary resources and painting a narrative of what life was like during that period. We also explored the cults of personality developed by the emperors and how they crafted their narratives for future generations.
On the business side, some of my marketing classes were absolutely intriguing. I did a lot of work in the marketing department at Penn and found the qualitative and quantitative approaches to understanding marketing strategies to be very interesting. Although the focus was primarily on brand marketing, there was also some insight into performance marketing. However, the marketing landscape has shifted significantly with the rise of Google and Facebook in recent years.
In terms of engineering, I loved taking a class on semiconductor physics. This class provided a perspective on how deep material science work powered computers, phones, and other devices, starting with the quantum physics revolution in the early 1900s and continuing through the research conducted at Bell Labs, Fairchild, Intel, and other semiconductor companies. It was fascinating to learn about the history and science behind this evolution.
On His Internships
I intentionally pursued two distinct internships: one focused on engineering and the other on business.
At Innova Dynamics in San Francisco, I was exposed to fascinating technology that involved embedding silver nanowires into plastics to create flexible touchscreen displays. As a material scientist, I gained valuable research and lab experience, but ultimately realized that the field was not for me in the long term, despite my intellectual interest.
I wanted to explore a more business-oriented career path for my second internship. I landed a position in Morgan Stanley's tech investment banking group, which provided a great learning experience and insight into the industry. While investment bankers are generally incredibly smart and hard-working, I quickly discovered that the career path did not fit me well. I craved a steep learning curve and didn't feel I was growing much in that role.
Despite this, I found the work itself to be interesting. We worked on IPOs and M&A transactions and created decks to understand where things were trending for clients. This experience confirmed that I wanted to pursue a career in this direction, though I wasn't exactly sure what that would look like in the short or long term.
I often advise people to try as many things early in their careers. The point of internships isn't to set a perfect career trajectory but to explore and learn what you like and don't like. It's okay to say no to something that doesn't feel like a good fit.
On Pursuing A Career in Venture Capital
Honestly, I didn't know I wanted to do venture the whole time. It was actually a complete accident. I had no idea what venture was until I attended a Bessemer information session for undergraduate students interested in joining as a junior-level analyst. I thought the role was cool and decided to explore it further to see if it was worth pursuing over the long term.
As I went through the application process, I learned much about what it meant to be a junior investor at a firm like Bessemer, where they've been running this program for many years. Fortunately, I was offered the position, and from there, it was off to the races.
From the Bessemer perspective, I learned that they give you a ton of independence. In some ways, they drop you into a room and tell you to find interesting businesses and markets and help discover the next Pinterest, Twitch, or Twilio. As a fresh graduate with little practical experience, it was a ton of responsibility but also an incredible learning experience.
One of the things that makes Bessemer unique is its intellectual odyssey. We're very thesis-oriented and like to go deep into spaces to understand what's happening in the markets and make investments in the best companies, from seed to growth, all the way to the pre-IPO stage. We really enjoy doing the work and value open conversations about what's working and what isn't.
Lastly, the people you work with at Bessemer are genuinely nice and care about your success, which has been a huge help for me.
On Proving Value As A New Investor
Being present and an active listener are key traits for success. It may sound simple, especially in junior roles, but as you begin to develop your career and skillset, it's important to have a firm point of view based on doing the work and talking to users, customers, and others in the industry. This allows you to form a synthesized view of where the market is headed and where things are evolving.
As a venture capitalist, it's important to have a high-level perspective of how the landscape is shifting over time and be able to predict how it might shift over the next 10 years due to trends. Engaging with others in the data community, financial services industry, or other relevant industries is important to gain perspective and put it together in a cogent way.
Sharing this perspective with founders can help build a better relationship with them. It's easy to default to passive venture mode and focus solely on the numbers when raising funds, but having a full conversation about the importance of the product and why people approach it a certain way can help build a stronger, trusted relationship with the founder.
Ultimately, this business is about building solid and trusted relationships with the founders and stakeholders you work with. As a board member or investor, you're on a long-term journey with the CEO and founder to steer the ship to its final destination.
On Shaping His Investment Theses
It's a great question of where to spend time because you can spend time almost anywhere. From an investment perspective, most venture investors try to look at the trends that will shape the world for the next 10, 15, 20 years, or more. What macro shifts are happening that will create entirely new industries or opportunities for technology or other businesses to take advantage of?
For example, personal experience can also inform this perspective, as in the gaming space. I was a gamer and saw many exciting things happening, such as the rise of massively multiplayer online games that turned into true consumer social platforms and e-sports teams. The rise of Twitch viewership of eSports is also interesting to watch, as it could impact the sports media economy.
In spaces like software infrastructure and data infrastructure, there has been a massive shift to cloud-based services, starting with the launch of AWS, and now there is a massive explosion and growth of application software companies as well as the infrastructure layer that powers many of those companies and other businesses that aren't traditional technology companies. All these businesses have technical teams that are starting to implement technology applications in the front end to consumers or use technology in the back end behind closed doors to power their businesses. This massive wave and shift have created an interesting opportunity for investors to focus on enabling those professionals to be more successful.
On Working at Viagogo
I left Bessemer to try something different at a young age, just to learn something new. Earlier in the conversation, we discussed trying new things to discover your likes and dislikes.
At Viagogo, my role was quite broad. I did everything from running payroll to helping close the books at month-end and eventually ended up in the product organization. We primarily worked on our supply team, finding people who were selling tickets and helping them sell more effectively.
It turned out to be more of a data science and data analyst role than a product role. We did classic BI reporting; most product managers were fluent in SQL. The organization valued people who could derive insights from the data highly. We also did things like experimentation and data labeling in-house, which I didn't fully appreciate then but turned out to be incredibly valuable to the business. We pushed in one constant direction to excel at data in the organization, which turned out to be a long-term differentiator and helped the company grow to massive amounts of revenue.
This experience exposed me to the world of software and data infrastructure, and I realized that even less sophisticated techniques can create valuable businesses over the long term.
I enjoyed my time at Viagogo and learned a lot, but missed the venture experience. That was the main thing for me - the ability to learn about new spaces and talk to people doing incredibly interesting things every day. It's a special thing to partner with amazing folks creating something from nothing and doing truly world-changing things.
That brought me back to Bessemer and keeps me going every day - the opportunity to see how the technology landscape is evolving and how those folks are powering it.
On Developer Platforms Investments
We have an entire article about the eight laws we look for, including on developer platforms.
One law, in particular, stands out as the most important: user love. It refers to love from the community or developer advocacy, which can sometimes be quantified through usage metrics, such as adoption and increased usage over time. It can also be qualitative, such as user feedback posted on Twitter.
When users gush about a product, it's a strong indicator of success. This is why LaunchDarkly and PagerDuty are such strong trends in the developer platform space; they empower developers to do their work better. Other Bessemer portfolio companies, like Twilio, SendGrid, and HashiCorp, also benefit from this trend.
On Data Infrastructure Investments
I’ll start with the teams. All three teams have deep expertise in the data space, with Matt from Coiled, Jeremiah from Prefect, and Gary and Raj from Arcion possessing extensive knowledge of the subject. We look for a long history of expertise and empathy for the user, which we found in the founding teams. All three teams had this trait, which was informed by their experiences working for companies that dealt with data problems at scale.
We looked for products that brought about a significant change in the way we work. For example, Arcion’s Change Data Capture offered a streaming and real-time solution, while Prefect proved to be a better option for Airflow’s DAGs that kept breaking. We loved using Prefect and saw improvements in our data. Both the team and product followed some of our developer laws.
On Education and Community Investments
Let's start with Tribe, which is a community management platform. It helps creators, SaaS businesses, and larger enterprises manage their communities.
Even before the pandemic, we saw a trend of solopreneurs creating interesting online businesses. They could reach an infinite number of people on the internet and find their niche community that was willing to provide them with an audience they could monetize and create a standalone business around. These were podcasters, e-commerce businesses that arose during the rise of Facebook advertising, and companies like Shopify, where we were early investors. We saw a continuation of this theme in various industries, with the community being one where we see the podcaster, the e-commerce brand, and Glossier, which started with a solid community of followers.
On the Guild side, there are multiple reasons for our investment. It is a corporate education-as-a-service platform that partners with companies like Disney and Walmart to provide their employees with education as a benefit. Employees can take classes and earn accredited degrees, which is an incredible proposition for both the employee and the employer who wants to retain and upskill their staff.
The CEO, Rachel, is a force of nature, and her vision and what she is building is truly remarkable. Although our investment does not have an explicit core focus area, we were compelled to invest because of Rachel's incredible vision.
On Advice for His Portfolio Companies
Navigating hard decisions and go-to-market strategies are two separate questions. On the go-to-market side, listening to your customers is crucial. Investors tend to over-rotate on particular good market strategies, and there is currently an obsession with PLG (product-led growth) and bottoms-up adoption. While this sometimes works, it may not suit every industry or segment.
It's essential to figure out who the user and budget holder are, as these two things may not necessarily be the same. In the data space, for example, data analysts and data scientists may not have the budget that developers do. Budgets may also live with someone else in the organization who isn't the explicit user. However, the data scientists and analysts may still have a big say in the decision-making process.
Making hard decisions is entirely varied and depends on the company. Every company has had to make a hard decision at some point, such as deciding whether to go out for a fundraising round or make a key executive hire. These are all tough decisions to make. The only universal advice is to have honest and open conversations from a board to a founder perspective.
Encouraging portfolio companies to plan ahead for different scenarios that may not be right around the corner, like a recession, is crucial. Being prepared and having open conversations about how to be prepared is necessary.
On Supporting Bessemer's Founders
Providing a high-level overview of Bessemer’s services may take some time. Much of the information is public, including benchmarking and deriving work related to data infrastructure, developer platforms, roadmaps, and scaling from 1 to 100 million ARR benchmarks. Our growth team, led by our partner Mary, has done excellent work in benchmarking across the portfolio, which we share with our portfolio companies to help them optimize their performance.
We also have an operating advisor network composed of VPs of product and CFOs in the Bessemer network who have experienced the challenges portfolio companies go through, from founding to IPO. They provide guidance and support on various aspects of a company’s life cycle, from making early hires to preparing for an IPO.
In addition to these services, we also organize community events to connect founders with others who are going through similar experiences, as well as assist with hiring for key executive positions like VP of sales, VP of marketing, or CFO to accelerate business growth.
On The Data Infrastructure Roadmap
Many of your listeners are likely familiar with the market trends I will discuss, so I'll keep this brief. We identified four key drivers for the development of what some are calling the modern data stack.
The first is the growth and adoption of cloud software. Companies of all sizes and industries are adopting cloud-based software to run their businesses, resulting in data sprawled across various systems like Salesforce, HubSpot, Desk, and dozens of other SaaS applications or databases. This led to the development of cloud-based data warehouses like Snowflake, Redshift, or BigQuery.
Startups are taking advantage of these flexible architectures and building products that serve data teams or professionals who want access to data from various sources and join them together and query them in interesting ways.
The second driver is the increase in the volume of accessible data. Reliable connections to these data sources become more important as more users generate more data.
Third, data is becoming more of a differentiating feature for many businesses, like the new oil. TikTok and Netflix have invested heavily in their data stacks to do things like personalized content and help with automated decision-making.
Finally, there is a huge demand for talent in the data space, resulting in a growing gap between supply and demand. Each incremental data scientist or analyst must be more productive, which comes down to having better technology to work faster and better.
All four factors excite us about investing in new products and companies that bring the modern data stack to life.
On Guiding Principles For Best-In-Class Data Platforms
We discuss four core guiding principles that we see in best-in-class data infrastructure platforms, some of which overlap with our ideas for developer platforms. One of these principles is user love, a fundamental aspect of all such platforms. On the ecosystem side, companies like Fivetran have done a great job partnering with Looker and Snowflake, which has helped drive their adoption.
When we talk to data practitioners, we find that they prefer products that are opinionated and well-integrated into the ecosystem of tools they use or want to use. Fivetran has done a great job of understanding and navigating this, which is a big reason they have become so successful.
Developing a community around your platform has also become increasingly important. dbt is the prime example of this, having created an incredible community that facilitates learning and helps educate people about how to do data work better.
Other important things include removing friction and enabling collaboration, which can be achieved by tools like HuggingFace in the machine-learning world or cloud-based notebooks like Hex. By working together, rather than being siloed away from each other, data scientists and other stakeholders can do their work more efficiently and effectively. This multidisciplinary approach is something that we value as we evaluate various platforms.
On Abstracting Away Complexity From Data Engineering Problems
Many people talk about data engineering problems that are incredibly complex, take up a lot of time, and are critical to the business. Just getting data from Salesforce to Snowflake has historically been difficult. There are platforms like Fivetran in the ETL space, and then there are reverse ELT tools like Hightouch or Census that pull data back into core operational stores like Salesforce, Hubspot, and Zendesk. These two categories are pretty interesting.
In addition, workflow orchestration tools like Airflow, Prefect, and Dagster are another core area in which we have explored and invested. We specifically invested in Prefect on the orchestration side because we continued to hear from more mature data organizations as they underwent their data transformation journeys that having a platform to help orchestrate jobs in a cloud-native way was becoming increasingly important. Prefect helps facilitate that, allowing data scientists to do interesting, forward-looking modeling in addition to traditional, backward-looking work.
On Powering The Next Generation of Data Scientists
This is a high-level view of the situation: As data volumes continue to grow, many business users are working with and creating analyses. They need to be empowered in the same way that software developers are.
We are excited about products that help data scientists increase their job efficiency and efficacy. Some of these products include tools for transforming data, such as analytics engineering with tools like dbt, and query processing engines like Starburst or Trino. Other tools include IDEs and notebooks that help with collaboration, like Hex, Notable, Deepnote, or Ponder. There are also larger data science platforms that encompass a variety of things, from pipelines to storage to processing, such as Databricks or Dataiku. Our portfolio company, Coiled, is also starting to do some of that in the Dask space, helping users parallelize their workflows in the Python native tools and environments they already know and love.
On The Emergence of Metadata Management
In terms of which area I'm most bullish on, I think a lot is happening in governance and management in various ways. So perhaps I'm cheating and saying both are actually pretty interesting, in my opinion.
On the governance and management side, there are data catalogs, privacy tools, lineage tools, and monitoring approaches.
On the monitoring side, different approaches are being taken, such as Monte Carlo and Great Expectations, with Datafold being another example.
Both areas are interesting, and we'll see a lot of consolidation of feature sets in both.
When you listen to users talking about their data catalogs, privacy, access tools, lineage tools, and observability features, it seems like they want a full suite of tools rather than just one thing. Folks are trying to figure out which wedge is the most interesting. In the governance space in particular, there's a really strong wedge to be had, given that access to data is a complex and essential pain point that seems to be becoming more important.
We have a portfolio company called Okera that helps enterprises organize and manage their data, and they're riding that tailwind.
On The Evolution of ML Infrastructure
Many of the themes in ML infrastructure are similar to those in data infrastructure, with slight differences. The traditional data analytics market is historically based on reporting past events, such as revenue for a specific set of customers over the past 12 months and marketing spending. However, companies that have adopted cloud-based data stores and software are starting to get excited about the potential of having all their data in the cloud. With this, they can make decisions based on the data, such as predicting customer lifetime value or churn.
Leading companies like Uber, Meta, Netflix, Bytance, and Airbnb have built tools to harness this data, but these tools are not necessarily available to startups or medium-sized businesses. This presents an opportunity for machine learning teams at these companies to enable themselves to do this work. We see three areas emerging as a result: data labeling, models as a service, and new databases.
Data labeling is vital for any machine learning work, and the need for it will continue to increase.
Models as a service, such as Hugging Face, provide pre-trained models that can be fine-tuned based on specific use cases.
New databases like Vector Database allow for the processing of unstructured data and can store relevant features, enabling faster search and recommendation systems.
The need for standardization is a critical aspect of the conversation around data and machine learning. Many heads of machine learning express frustration with the proliferation of point solutions, making it difficult to establish a canonical stack for machine learning. Standardization is necessary for everyone to agree on what tools to use for specific use cases.
Open-source projects from companies like Airbnb, Uber, and Netflix are still in the early stages of commercial viability for startups and medium-sized businesses. Making these tools work outside of the specific infrastructures of these companies requires deep technical infrastructure work.
These challenges are not new to the industry. Bessemer's focus on developer platforms in its investments, such as Twilio in 2008, showed the second empowerment of all technical and business buyers. This led to the product-led growth revolution and motion, which expanded beyond developers to marketers, customer success, data science, data engineering professionals, and machine learning practitioners. PagerDuty, HashiCorp, LaunchDarkly, Coiled, Prefect, Arcion, and machine learning tool platforms are examples of this expansion.
On The Next Wave of BI and Data Analytics Software
There are three major trends in the field of analytics. Firstly, there is a shift from batch processing to real-time data processing, as people demand data with much lower latency. Platforms such as Imply (built on Apache Druid) or Startree (built on Apache Pinot) provide OLAP databases and front-ends that enable developers to deliver real-time analytics to end-users.
Secondly, there is a growing emphasis on augmented analytics, which involves automating the analytics process for tasks like root cause analysis and triaging incidents. Tools such as Sisu Data can help with this.
Lastly, there is a need for vertical-specific applications that provide tooling to understand and work with data for industries that have not yet adopted data infrastructure explicitly. Companies like Optimal Dynamics and Syndio are helping trucking logistics and pay equity analysis, respectively, to optimize their processes and understand their data.
These trends reflect a growing demand for more efficient and effective use of data in different industries.
On Investing in Climate Change and Student Builders
As I mentioned earlier, we at Bessemer like to focus on trends that are shifting the world and what they might enable. These two areas are informed by our pursuit of understanding what is happening today that might impact how we consume energy, content, or education over the next 10 to 20 years.
It's clear to me and my colleagues that a lot is happening in the energy sector, specifically the massive transition to renewables. Although it currently only represents a fraction of energy generation, it will grow to a much larger portion over the next 20 to 30 years. As we put solar on roofs, wind on the ground or offshore, and have batteries and electric vehicles running around on the grid or in homes, we'll witness many other transitions that will change how energy has been consumed and generated for decades, maybe even more. This shift will also result in many interesting businesses, whether they are deeply technical, like material types of businesses, software businesses, or SMBs that put solar on roofs.
In terms of education, the explosion of content online on platforms like YouTube or Coursera has allowed individuals to educate themselves, share their work, and help others learn in a more community-oriented way. This is particularly exciting as we witness a shift in how education will be approached in the long term, even though traditional educational institutions are not going anywhere for quite some time.
Show Notes
(01:40) Sakib shared formative experiences of his upbringing in SoCal and his undergraduate experience at the University of Pennsylvania.
(05:24) Sakib recalled his favorite classes at Penn.
(07:55) Sakib reflected on his internship experience at Innova Dynamics and Morgan Stanley.
(11:04) Sakib reflected on his decision to pursue a career in venture capital at Bessemer Venture Partners.
(14:02) Sakib walked through his process of proving value as a new investor.
(16:21) Sakib explained his process of forming clear investment theses.
(18:35) Sakib talked about his brief year as a product manager at Viagogo before going back to Bessemer.
(22:04) Sakib dissected his investments in LaunchDarkly and PagerDuty (in the domain of developer-centric platforms).
(24:16) Sakib explained his investments in Coiled, Prefect, and Arcion Labs (in the domain of data infrastructure).
(25:59) Sakib walked through his investment in Guild Education and Tribe (in the domain of education and community management).
(29:06) Sakib shared advice to portfolio companies in terms of navigating hard decisions and growth strategy.
(34:18) Sakib outlined Bessemer's roadmap on data infrastructure - which looks at the wave of startups enabling the next generation of data-driven businesses.
(39:57) Sakib brought up the products that help abstract away complexity from data engineering problems.
(41:34) Sakib highlighted the tools that power the next generation of data scientists.
(43:27) Sakib emphasized the emergence and evolution of metadata management
(46:48) Sakib unpacked the evolution of ML infrastructure.
(49:28) Sakib examined the key trends and opportunities that will define the next wave of BI and data analytics software.
(52:56) Sakib shared his investment perspectives on climate change and student builders.
(55:29) Closing segment.
Sakib's Contact Info
Mentioned Content
People
Sarah Catanzaro (General Partner of Amplify Partners)
Ed Sim (Founder of Boldstart Ventures)
Mike Speiser (Managing Partner of Sutter Hill Ventures)
Book
"The Idea Factory" (by Jon Gertner)
Notes
My conversation with Sakib was recorded back in late 2022. Since then, I recommend checking out these resources:
This blog post on the era of intelligent search
Bessermer's AI Roadmap and the ChatBVP bot
Bessemer's 2023 Cloud 100 Benchmarks Report
About the show
Datacast features long-form, in-depth conversations with practitioners and researchers in the data community to walk through their professional journeys and unpack the lessons learned along the way. I invite guests coming from a wide range of career paths — from scientists and analysts to founders and investors — to analyze the case for using data in the real world and extract their mental models (“the WHY and the HOW”) behind their pursuits. Hopefully, these conversations can serve as valuable tools for early-stage data professionals as they navigate their own careers in the exciting data universe.
Datacast is produced and edited byJames Le. For inquiries about sponsoring the podcast, email khanhle.1013@gmail.com.
Subscribe by searching for Datacast wherever you get podcasts, or click one of the links below:
If you’re new, see the podcast homepage for the most recent episodes to listen to, or browse the full guest list.
