Twelve years into my career (Linkedin), I’ve collected many little lessons and stories I wanted to write down (see a prequel blog post for stories and lessons prior to graduating college and a sequel including my career at Square).
The Federal Reserve Board (DC) – Research Assistant (2013-2015)
My first two years after college (2013-2015) were at the Federal Reserve Board in Washington DC in the Macroeconomic section, specifically the Price and Wages group (i.e. inflation forecasting) as a Research Assistant (RA). There I learned how to code, use unix, and manage a database as well as work a 9 to 5 job in an office environment. The coding was mostly in FAME (I started having dreams about FAME that looked like the matrix after a few weeks) and RATS (I was fortunate to have college experience in RATS from an obscure economics class which helped me get the job).
The Fed had just surpassed a century since it’s founding when I joined, so some programs had code comments as early as the mid-90s. Someone I worked with had been forecasting food and energy prices since before the internet. While the code was old, the economists were very, very strong at forecasting. They knew the data unbelievably well, and were very good at not overreacting to new information in adjusting their forecasts. It was run incredibly well and I had a hard time finding ways to improve things as it was so focused and mission-driven with established processes and expertise. My second year I spent a lot of time cleaning up old code/tech debt for fun (it was very satisfying and gave me a sense of what was actually really important in the code)
This is where I learned to understand an area so well that it became almost boringly predictable (I also learned I didn’t find academic research very engaging, it felt too far from people – though I was fortunate to partner on two FEDS notes: one and two). I also learned how to get things exactly right with formulas – rather than falling back to guesswork/ballpark figures I had done with homework, etc in school when I wasn’t sure (in retrospect, coding taught me logic much more directly than math problems / proofs). There were clear RA instructions for a lot of key production events (like monthly data releases from the BLS and BEA), and I enjoyed improving the wording and finding ways to set things up so they could not fail (e.g. in a program: first make a copy of a database, then run the update commands, and then overwrite the original – so if there was a failure during the update commands we wouldn’t need to re-create the database). Little did I know I was also learning about a different form of git, production vs staging environments, and idempotent operations.
I began to realize that part of the reason for forecasting was to develop a hypothesis of what to expect, and then when the actual data came in you could see how close you were and what contributed to the differences with reality. I re-built some time series forecast models in R to learn that code (and understand the models more deeply), but really after two years I was ready for something different than backend pipelines and research assistant data work. In retrospect being an RA was almost like getting a partial economics masters (many research assistants went on to grad school) and learning how to understand models, write code, make charts, etc. My time at the Fed was probably when I was the strongest technically in my career and though it was really transferrable to ML I felt so distant from the impact and tech industry practices that it took me a while to realize how much I had actually learned at the Fed.
As a cool aside, I delivered inflation packets every month to Ben Bernanke and later Janet Yellen in their offices. I was also a fact checker for the Monetary Policy Report and was once on CSPAN (in the background). The work was maybe a bit boring after a while and felt too far from feeling an impact directly on the broader world (granted some of that might have in retrospect been me struggling with no longer being an athlete or a college student and just being a white collar worker), but it had some cool perks and was an excellent transition to the working world after college. I appreciated the work-life balance and working with so many brilliant economists/PhDs. It made me realize that no matter how smart / capable people can be – no one knows the future, at best we can know as much as we can about the present and choose where we focus.
Side note: during this time I also learned a lot about finance outside of work. I got my first bank account, credit card, 401k, started student loan payments, paying rent, etc. I felt like a real adult when I first started doing my own taxes.
Moving From Washington DC to Sonoma County
In 2015 I got married (in DC) to the love of my life and we moved to California so my spouse could run a fashion jewelry store in her hometown of Windsor, CA. I moved to California without a job lined up and was very fortunate SoFi had an operations call center 10 minutes away in Healdsburg.
SoFi (Social Finance) – Data Analyst (2015-2017)
Where the Fed was established and predictable, SoFi in 2015 was hacky and chaotic. I joined SoFi the month after they raised $1B from SoftBank – the largest single financing round in fintech at that point. I think the company had doubled in size every year for several years and was continuing to do so, which invited a lot of change and chaos. It was a wild time to be there.
The original idea of SoFi was pretty clever. The founders noticed that Stanford MBA students had the typical ~6-7% interest of most student loans – BUT they hadn’t had a default in 30 years – so they could definitely pay lower interest without much risk (and with low interest rates broadly, getting financing was relatively cheap). The founders raised a few million dollars and went out to refinance these student loans – however, when they approached the MBA students they were like ‘I’m sorry, who are you?’ and few actually refinanced. This is where it became apparent that a solid brand was also a prerequisite to a Fintech company’s success – and that took marketing messaging and repetition (and some community focus, especially early on). They had since expanded to personal loans, mortgages, and wealth management. The emphasis on growth meant they kept needing to add products and find big markets to justify the valuation, etc.
My first project was effectively building a database in an excel spreadsheet. I hacked it together and learned a lot about excel functions (vlookups, pivots, etc). However, a month later the VP who led the project was fired and my work was thrown in the garbage. I thought this was maybe just how to private sector operated (since I had only worked at the Fed), and I focused on learning what I could rather than getting attached to any specific project. I spent my first few months on creating operational reports with vendor data, later automating it in R (since it was free and I was familiar with it) – building out what were effectively hacky ETL pipelines of daily programs to be run to show worker efficiency, loan review numbers, etc. Automating all these reports (I added up the time) saved the team of 3-4 analysts about 4 hours of manual work a day – but my skip lead wasn’t very technical and didn’t see the value of it (he was more into sales / loan numbers than technical stuff). I began to grow frustrated with the organization and started feeling under appreciated by management (I just remember feeling angry all the time). The business operations analysts would also frequently get interrupted with increasingly urgent questions from a broad range of stakeholders – sometimes being told ‘work on this instead, it’s more important’ mid-way through working on something up to ten times a day. It became hard to focus and I needed earplugs and noise cancelling headphones and sometimes to hide in our open office plan/call center to focus and get things done. My favorite example of focus ‘defense’ was how IT would have a white board display of what they were working on, ranked by priority (separate from Jira), so when someone came in screaming for them to fix their problem they would say ‘okay will do, what should I take down? Fixing the CEO’s laptop?’ which would help that person back off. Everyone was pretty stressed and constantly putting out fires while navigating a lot of changes due to the rapid growth and high turnover.
For context on what this much growth was like, a few months before I joined, they ran out of room for people to fit in the building. As a result, while they were setting up a lease next door from their Healdsburg office, they split into a day shift (6am-3pm) and a night shift (3pm – midnight). However, management worked a more typical 8am to 6pm, so from 6pm to midnight there was very little managerial oversight – and IMO this lack of oversight grew to a lot of toxicity, fraternity culture, and future lawsuits (article) – but honestly who knows.
Leadership at the time often prioritized finance, engineering, and marketing – ie raising money, over actual operations of processing loans and customer service. I learned over time that business cultures often follow similar prioritization, where some business areas (like customer service, IT) are seen more as ‘pass-fail,’ meaning unless they are clearly broken, they are not improved or invested in (i.e. a C grade is the same as an A to leadership). Meanwhile some areas of the business (often product or marketing) are areas of growth where they want to get the best organization they can get (i.e. leadership wants to go for an A+ and will spend more money to get the best people, culture, etc). It’s a tired story now, of hyper growth obsession in tech, but at the time I was learning what it looked like in practice / being there. I focused on my area of the data and tried to stay out of the chaos as much as I could – but I observed and learned the whole time.
My technical strengths gave me a seat at the table for higher level decisions, because I was closes to the data and information (which bottlenecked since database access was largly restricted for security purposes). Because of this I got to work on all sorts of problems such as email, call, chat, financial, compliance, marketing data, etc (even effectively doing payroll at one point). My favorite example of which was working with operations and marketing to clarify loan start and submit definitions that effectively ended months of arguing: marketing was getting people in the door and saying ‘okay, we are done’ (application starts / soft credit pulls), while operations was waiting in the living room to process loan applications saying ‘where is everybody?’ (hard credit pulls) and I helped explain that there was a hallway between the two (i.e. more clearly defining the funnel conversion steps), and that they were using two different definitions for the same word: ‘submits.’ Essentially Marketing went from A to B, Operations went from B to C, and I clarified definitions and laid out the A to B to C steps and updated a daily snapshot of loan progress for the company and helped create a clearer source of truth, reducing those arguments to zero (i.e. I remember afterwards the CEO saying to the head of Marketing: ‘I don’t care about soft credit pulls, I care about hard credit pulls’). I felt so proud of the clarity I brought in – seeing it as preventing future arguments and confusion.
In the startup environment, I appreciate that I learned APPLIED data analysis, where I worked as a data service across the entire company on improving decision-making, campaign targeting, credit risk assessment, compliance, etc. It gave me broad exposure and taught me a lot of working people skills and empathy with each part of the business to understand their incentives and what goals they were trying to achieve. A small change in context or a little more domain knowledge for a request could make the work MUCH harder or easier. I started to see all incoming request as contracts of which I could clarify, negotiation, and prioritize before agreeing to them (whereas initially I felt obligated to do anything that was asked of me).
At one point in my tenure at SoFi I found out my salary was the second lowest at the company. After that I tried to get a raise several times and failed (it led to some awkward chats with my boss where I knew I was underpaid, but couldnt say how I knew because it was shared with me on accident). My only leverage was my marketability; the best strategy for advancement was to leave and get a better paying job elsewhere. There became a ‘death spiral’ where the most organized and effective people got surpassed by the more aggressive and loud people, so they left, making the operations org more chaotic.
With all the chaos and frustration at SoFi, I probably would have left much sooner if I had more options. I genuinely loved the work with data, but the culture in operations at the time was terrible (loud & inconsistent, meaning I often had to working late and on weekends to get impactful things done). But I lived in a rural area (wine country) and the only other interview I could get was with the local sheriffs office as a financial analyst. I felt stuck (which later motivated me to move down to the SF Bay area with my spouse, where there were so many more options). I was fortunately able to navigate transferring to engineering after a year at SoFi where I learned deeper and more transferring technical skills (i.e. going from excel to SQL and Tableau) and the culture was much better. I worked on more interesting problems than just chaotic operations (e.g. credit, marketing, product, finance), and got to work from home most days – commuting 1 day a week from Sonoma County to the SF office (I would leave at 430am to beat traffic). I effectively went from working in a crowded office space on top of a restaurant in Healdsburg to working in a classically fancy tech office in the Presidio in SF – the transition was mind-blowing. It felt similar to my transition from a large public high school to a fancy small private college. My second year at SoFi passed much more smoothly.
Interviewing
At some point I realized that while things were improving at SoFi, I could just move laterally to another company and have a much better salary and overall culture (i.e. clear up some of the resentment and baggage I had built up). I also realized after failing a random analyst screen from Facebook that maybe I couldn’t get my own job at this point if I applied to it. I needed to learn how to pass tech interviews. My interview at SoFi had been relatively short and fast (in retrospect it was a red flag that everyone interviewing me seemed so exhausted or indifferent). My interview at the Fed was from undergrad and more formalized. I wanted to learn how to interview at a more prestigious tech company. I reached out to my network from college, got referrals, applied to jobs (see my blog post on a simple job application strategy), got through SQL tech screens, and kept a healthy pipeline of job interviews until I got some on-site interviews at Facebook, Google, Instacart, Square, and Blend. I tried to set up a ‘funnel’ of job applications where I would try not to be too busy or exhausted, but I always had a next step for moving through interviews (e.g. finish a technical screen, start trying to get another one set up, etc). I learned more of my weaknesses in data (for example Instacart told me they didn’t hire me because my inferential statistics skills were weak, so I took an online course on inferential statistics as a refresher, and a few weeks later Google told me my statistics skills were strong). Anyways, I ended up at Square in late 2017. While SoFi’s main product at the time were loans (i.e. FIN part of FinTech), Square’s origin story involved smartphones, a card reader, and payments rails (i.e. hardware and software, the TECH part of FinTech).
Part of the reason I liked these tech companies was (1) I could relocate to the SF Bay area, so I wouldn’t get as stuck as I had felt at SoFi in Healdsburg, and (2) when people LEFT those companies they moved to cool startups or other cool tech companies, whereas when my peers left SoFi they often were quitting out of frustration and working at smaller/older financial institutions. Tech seemed like a more nerdy and fun environment than pure finance at the time, so I wanted to ‘invest’ in building a stronger network there. What really kicked off my search though (besides my spouse also being ready to move on from retail), was when I went to a retirement party in DC for a former manager at the FED – I felt so respected and proud of having worked there, which made me realize how frustrated I had become at SoFi. Remembering where I came from was the perspective I needed to take action.
See the next chapter in my career at Square! Or previous chapter on odd jobs before graduating college.
Daeus Jorento 