My first big girl job outside of college was FloodNet. Having now been here for far more than 6-months, I feel a sense of intrinsic devotion that I haven’t felt for a company before. This isn’t to say that my previous experiences prior to FloodNet meant nothing, because they absolutely did. But there is only so much you are willing to put into a place you are only going to be at for a brief period of time. Now in my second year here, I am able to look back and proudly see how far both me and the project have come. I have ownership over several elements of the project, a great degree of independence, and have pleasantly found myself with mentees. There is a lot that both FloodNet and I have accomplished over the last few years, and there is still a long way to go. Thank god, both the project and I have absolutely no intentions of slowing down anytime soon.
While my offer being rescinded from WHOOP was devastating, it did mean that I was finally free as a bird to do whatever next my heart desired. However, during the summer of 2022, there were massive waves of layoffs throughout the tech industry. People with countless years of experience at well renowned companies were rendered jobless, and here I was a fresh graduate out of college competing with them. The air of uncertainty in the job market made my search process arduous and quite disheartening at times. Despite everything, I knew that there were two things that I was unwilling to compromise on no matter what:
This is when I happened to stumble upon FloodNet.
To be honest, describing the origins of FloodNet to most people is a complicated task. It is not helped by the fact that I myself have a loose understanding about how it all began. From what I do understand though, a professor at NYU specializing in wastewater treatment encountered a roadblock during her research: there was no quantitative data on flooding in NYC. Yes, we are all aware that New York, like most cities in America and quite honestly the world, floods. It has since the beginning of time. Yet what we don’t know is where in the city it floods, how often it does, what the rate of onset is, the depth, etc. Motivated to correct this oversight, a low cost sensor that provides real time flooding data was designed in a combined effort by researchers from NYU and CUNY. Fueled by the devastating aftermath of 2021’s Hurricane Ida, many community members and government agencies have since gotten involved with the project. What was once a mini research project starting in a lab at NYU is now a multi-million dollar city funded initiative. The goal is to have a total of 500 sensors deployed across all 5-boroughs of NYC by the end of 2027.
When I came across FloodNet I saw a great opportunity to not only grow as an engineer, but to also give back to the people of NYC. You see, for me New York isn’t just any city on the map. It is the city that has given countless immigrant families like mine opportunities. It is the city that my family chose to immigrate to and has spent more years than I have been alive building a name for themselves in. It is the city I was born in. It is my home. So when I applied to FloodNet it was in the hopes of making a lasting impact on the city that has given both me and my family everything.
I originally applied for the role of a Firmware Engineer on the project. Given my two firmware related co-op experiences, I thought that I had already pigeonholed myself for positions only in firmware. Luckily FloodNet is a company that puts in a lot of thought and effort into each application they receive. While I did not fit the requirements of the Firmware Engineering position, I was urged by the tech lead to apply for the role of the Sensor Engineer instead. I am not going to lie, I was extremely hesitant at first. All my previous professional experiences were typical 9-5 desk jobs. Even though I was used to working hands-on with hardware, I certainly had no experience climbing up ladders PSEG style to deploy sensors. I had spent very limited time in a lab using power tools. It seemed to me that the role of the Sensor Engineer required me to step out of my comfort zone, and I was not sure that I was ready for such a drastic change. What convinced me in the end, however, was an opportunity to be a part of this once in a lifetime project. In the heat of Summer 2022, I must have applied to hundreds of jobs at mediocre companies. Most of them blurred together at some point, but I specifically remember FloodNet standing out to me and getting excited. I knew I had nothing to lose by at least applying for this position, so I ended up taking a leap of faith. And as it has turned out that while I didn’t lose anything, I have gained more than I could have ever imagined.
Going in all I knew was that as a Sensor Engineer I would be responsible for the build and deployment of flood sensors. The exact processes or just how responsible I would be was up in the air. You see all my previous experiences were at bigger more established companies where there were hierarchical ladders. I reported to my specific manager who in turn had their own managers and so forth. My roles were always well structured and I had set deliverables to complete in a given time period. There was never any confusion as to what I should be doing at any given point of time. But FloodNet is not your usual run of the mill company.
First and foremost, everyone on the team is an academic or researcher. When I worked in consumer electronics, a product was shipped out on a specific day whether or not it was fully ready. There was a hard deadline that was set and honored. At FloodNet the quality of the actual technology and how it is being used is more vital than anything else, even deadlines. This results in more times than not, scheduled deployment trips being pushed back (or up). Trips have been postponed for a variety of reasons from the latest firmware not being ready on time to running out of an important component on the sensor. Trips have also been moved up because we were able to, albeit painfully, refurbish missing components from old sensors. Working at FloodNet means having the flexibility to adapt to very last minute (emphasis on the very) changes. You can spend days working on a task, but you have to always be prepared to abandon it at any given time as priorities shift.
The one thing that I would like to highlight is that in all my time at FloodNet, a deployment trip has yet to be rescheduled due to poor weather conditions. We have gone out in the field in extreme sunshine, downpours, and even when the sky was orange from the Canadian wildfires. This fact alone speaks to the resilience needed to work at FloodNet.
When I joined FloodNet I was the fourth person to come onto the tech team. Having just received funding from the DEP, FloodNet was slowly starting to expand. The network itself stood at a mere 30 sensors. As a later addition to the core team (everyone else had been a part of the project since day 1), a clearly defined path or documentation to get me up to speed simply did not exist. I was just thrown into the deep end and expected to figure things out on my own.
Within my first week of employment I was elbows deep into the v2.5 sensor builds. By the end of my first month I had built at least 100 v2.5 sensors from scratch! Before I joined the team there was only one version of the sensor to exist - the v2.0. The v2.5 and v.2.0 are both LoRa based devices. This means that in order for a sensor to function a gateway must first be set up within a mile range. The rapid scale of production was due to the fact that the v2.5 was supposed to be the final sensor version (lol but more on that later) that would be deployed all across the city. Most of its changes were minor design upgrades from the v2.0 i.e. the switch now being accessible from the outside.
Building the sensors from scratch involves a series of small intricate steps. Before actually assembling the sensor, individual components, such as the Maxbotix and LED switch, have to be manually soldered to their connectors. Now I had done some soldering during my time at Northeastern, but it was never that deep. Worst case I would damage a $5 component that was easily replaceable. I was also not mindlessly soldering for hours straight in a lab, making one more prone to slipping up. However at FloodNet, the stakes are much higher. Messing up and accidentally burning a $100 Maxbotix was a costly mistake that needed to be accounted for and explained to the city. There were consequences. On the bright side I am now a pro at salvaging parts from a bad solder job with a wick or heat gun. I have also since then dabbled in SMT soldering for the boards of later sensor versions. If I thought that inhaling solder smoke was bad, imagine my horror of sitting hunched under a microscope for hours tweezing after the world's smallest components.
One thing I love about the FloodNet sensors is how sleek and compact they are. The small size makes it incredibly easy to carry the sensors around while deploying them. Not to mention how chic they look out on the street.
But the sensor’s small size is also the one thing I detest about it. There is absolutely no wiggle room once everything is inside the housing. Every component is spaced out just enough to not be a hindrance to the other components. This makes it incredibly difficult to even put in anything in the first place (unless done in a very specific order and manner). Once the components are in, then there is no way to use the full range of a wrench to properly tighten them. As you can probably infer a lot of the build has to be done manually by hand.
Getting the sensing elements of the sensor together is just one step of the hardware build process. How the sensor will be mounted out on street-level is an equally important consideration. FloodNet has been given special permission by the DOT to use any/all drive rails throughout the city. But as we all probably already know, drive rails are not always in the best of conditions...
A sensor is currently mounted onto the drive rails via 1-foot aluminum struts that are then in turn attached to metal L-plates. The strut channels arrive at the lab as 3-feet. They are tapped with M6 holes using a CNC mill machine before being cut up into 1-feet increments with a saw. This whole process is done manually in the lab. The reason that we mount the sensor onto the strut is so that there is a range of motion to level the Maxbotix on the sensor relative to the ground. Having an L-plate on the aluminum strut allows us to move the strut if needed to counteract any inconsistencies and tilts in the drive rail.
The most prominent feature of the sensor is the solar panel mount perched atop the aluminum strut. The FloodNet sensors are entirely charged by solar energy, which adds to their independence and simplifies maintenance. Prior to me joining FloodNet, the mount that was in use, while getting the job done had several design flaws.
Our mounts have a 45-degree angle to them. This accommodates for the sun’s high position in the sky during the summertime and its low position during winter. In the design above, the panel is positioned over the screw hole to secure the mount to the strut. This made it difficult, and honestly quite dangerous, to use an electric screwdriver at an angle while up on a ladder to properly tighten the mount to the strut.
Another design flaw was the solar cable being directly soldered onto the solar mount at a 90-degree angle. The handling of sensors in the field can be quite rough, especially when working with numerous sensors simultaneously in harsh weather conditions and under time constraints. It is easy to overlook the need to handle the sensors with caution and end up tugging on the solar cables. There were a few incidents in the field where the tension from tugging caused the solar cables to detach from the panels.
At FloodNet, the goal isn’t to just get the job done. The goal is to get the job done as efficiently and effectively as possible. One of my first tasks after joining the team was to completely redesign the solar panel mount to address some of the design issues from the original version. After months of me struggling in Fusion360 and countless iterations later, the current design being used out in the field was adopted. The main changes from the original mount include:
The aluminum strut, metal L-plate, and solar panel mount along with any nuts/bolts needed are the bare necessities of a simple deployment. However, most deployments are anything but simple. Over the course of the last 2-years, at an ever increasing deployment rate, we’ve discovered a few fan favorite special components to have on hand. These include but are not limited to:
To maximize efficiency in the field it is important to be as bare-metal as possible when out there. This meant fine tuning the quantity and sizes of all nuts/bolts needed for each deployment on our end. All tapped holes (3 in total - two for the actual sensor and 1 for the solar panel mount) on the struts are M6s while the actual bolts needed to mount the struts onto the drive rails are M8s (2 bolts, 2 washers, and 2 nuts for a basic deployment). Additionally, it is also important to pack the correct tools. At the very least specific sized wrenches (all stop sign nuts can be loosened with a 9/16), an electric screwdriver, and a drill with certain socket attachments need to be packed. Oh! And don’t forget to bring a ladder.
For the most part we have a fool-proof plan and several backup plans down to a T. But I truly cannot emphasize enough how one can never predict what one will experience in the field on any given day. Some days are suspiciously too easy while others surprisingly difficult. There have been so many deployment situations that I simply could not have fathomed… While it is near impossible to proactively be on top of every single deployment situation, I have learned it is best to always be ready to think on your feet and be up for some sidewalk engineering.
Getting together the hardware and the means to put it out on the streets of NYC is only half the work that goes into the deployment of a sensor. While the amount of 500 sensors commissioned by the DEP may at first sound absurdly big, if you really think about it, it isn’t enough. Most of NYC is flood-prone. It is an island not to mention how the city was literally built over active waterways. There are just so many potential flood-prone locations across the city that sometimes it gets difficult to determine where and how many sensors to allocate in specific neighborhoods.
FloodNet takes feedback from residents, government agencies, etc. to determine potential locations to mount at to best capture floods. The field team then takes the locations to determine ideal drive rails to mount through elevation measurements in Cyclomedia, an online database that captures data from the real world using 360 degree street-level visualizations and panoramic photographs. Cyclomedia gives us the ability to see what drive rails are present out in the field and to determine the lowest point of watershed at a specific location. Potential drive rails along with their elevation values are scouted and saved in Cyclomedia as a JSON. A Python script is then run to convert it into a CSV file that can then be added on as a layer to a Google Map. The Google Map is then accessed in the field.
It is important to note that the locations we are given are contingent on there being viable infrastructure being present in order for us to mount there. What we seek in and around an ideal drive rail once we are at location is that:
As it turns out just because an area is easily accessible by floods does not equate to it being the same on public transit. This leaves driving as the best option in getting around the city for sensors. I have highlighted in the past that driving in Boston changes you for the worse, but that is nothing compared to the metamorphosis I have undergone driving in NYC. I wish I could speak on the things I have seen or done… A valuable life lesson I have learned, however, to prevent the mental breakdown one gets driving in NYC, is to also not follow the rules of the road.
Entering into my third year at FloodNet, I currently lead the charge on all field operations. With the sensor network having expanded dramatically during my time, I have become the go to person for all things field. Ultimately the FloodNet sensor network will consist of 500 sensors, making it the largest sensor network in the world. For such a big network, the number of people behind its success are not that many. This makes it easy for me to be in charge of many aspects of the project. I have at the very least been involved in the physical build, if not the deployment, of every single sensor deployed over the last few years. I am the one who upkeeps the current network through a handy dandy grafana dashboard I set up using SQL queries. Metrics tracked on this dashboard include: number of days a sensor has been down, sensors with a low battery trend, sensors with low LoRa network coverage, noisy sensors, and sensors with too many no return values. I am the one who then logs and tracks issues through Directus and GitHub.
Taking into account sensors that need to be maintained along with any new deployment requests, I am the one to then decide when and where a field trip is needed. I am the one to book a car through a local car rental place and map out the best possible route to hit the most down sensors. I also make sure that everything we can possibly need is packed efficiently the night before a field trip. When I say that the field is my thing, I truly mean the entire start to end process falls under my leadership and judgement calls.
As mentioned before, prior to FloodNet my roles at the other companies I have worked at were very well-defined. There was constant oversight over me with well defined deliverables. At FloodNet, however, I have an immense amount of freedom and independence to work on anything I want to. Of course as a Sensor Engineer hire, I still need to constantly be deploying and maintaining the sensor network. But my actual day to day and how I go about doing so is totally up to me. At first this was actually something I struggled with. It is so much easier to just be told what to do and have a clear deadline. Nevertheless, developing the skills to be self-motivated and trust my own gut have turned out to be more valuable in the long run. Everyone at FloodNet is overworked and constantly swamped with their own set of tasks. If I were to wait around for direction or help, I would get nowhere. Just the other day, I discovered that a bunch of sensors, specifically in Manhattan, were struggling to charge during the winter. The sun’s lower arch during the wintertime was making sensors surrounded by buildings less prone to charge, and ultimately led them to die. Instead of waiting around to get a consensus from the team for next steps, I decided to put a halt on all future deployments and instead prioritize possible solutions. I immediately mocked up solar panel mounts with varying angles instead of the standard 45 degrees, as well as a firmware code fix. So that by the time our weekly team meeting was here, not only did I introduce the problem to the wider team, but I also presented possible solutions currently underway.
While I am responsible for running the field operations, by no means could I have done it entirely on my own. Over the course of the last two years, I have had the unique opportunity to closely mentor and guide three team members. FloodNet likes to hire NYU Grad students to help out with the field work. At the very least, two people need to go out into the field for it to go smoothly. It is entirely up to me to interview, onboard, and get any new hires related to field operations up to speed. Prior to FloodNet, I had never envisioned myself ever being in a leadership position. It’s not that I thought that I would not have been good at it, but being responsible for someone else’s quality of work in addition to my own, seemed overwhelming. That all changed though when I found myself constantly struggling to single-handedly keep up with the demands of the field. Albeit painful to admit, I knew to succeed I needed to go against my entire being and ask for help. In the long run though it has all worked out for the best. These students have helped me out immensely with the physical demands of deployments, in addition to providing reliable new ideas and feedback. I have learned many new things from them that have helped better my own engineering. In exchange, I was able to provide them with meaningful career mentorship. The result of which is them all going on to have successful careers beyond their time at FloodNet.