I am a Software Engineer at Flatiron Health. You can also visit my Linkedin.

I returned to Flatiron Health for full time work. Over the last year, I have been working on data ingestion. Large datasets involving cancer patients need to be exhaustively cleaned in order to comply with HIPAA and protect patient privacy. I work on the singular dataset that other teams pull from to analyze. This involves a lot of engineering good solutions to processing large amounts of work. Additionally, I am often pulled to make performance improvements for other teams' workflows, for teams that specialize in medicine and not programming.

Between the summer of Junior and Senior year, I flew to New York City to intern for Flatiron Health, a biotech company working on using Data Science to solve cancer. I worked on a great, high-impact project trying to solve technical issues around data ingestion. To put it simply, medical data is often stored in PDF reports. This is a problem. When people are looking to see what pharmaceuticals are working or trying to gauge rates of diseases amongst demographics, they need structured data. To move PDFs to databases, abstractors with medical backgrounds are tasked with translating the documents. This is both time consuming and expensive. It would be much preferred to run an Optical Character Recognition (OCR) program on the PDF and extract the data that way. My project started out as a proof of concept, just to see if using Amazon Textract's Table function was even viable. As I moved forward, the pdf extraction was actually more accurate than the human readers. Also, it ran in less than an hour and cost a tiny fraction compared to the human readers. By the end of the summer, the scope had grown to adding many different types of reports, and building functions around them.

I worked for UC Davis as a student lecturer on ECS 98F: The Missing CS Quarter, which was inspired by a similar MIT course. Through my own internships and seeing my classmates, it was very clear that there were some skills that Software Engineers should have, but are not explicitly taught in core CS classes. For example, Regex, Git, Unit Testing, and Debugging are all units in this 10 week long class. I wrote homework assignments, held office hours, and given lectures for two years while an upperclassman at UC Davis. It was a valuable way to get more experience in these concepts, while also enhancing my soft skills.

I worked for my University, analyzing their environmental data. I worked split between the web development and data science team, which gave me a breadth of experience. I got the see the data pipeline from start to finish. For the data science team, I mostly focused on examining data and finding faults. Sometimes meters go down, and sometimes they are not properly aligned. The faster an anomaly is detected, the faster it can be fixed. I created a neural network model so the engineers could see detected anomalies with a User Interface. As a web developer intern, I also worked on multiple websites for the ECO team. There were internal facing websites, to show the board and managers that financial goals were being met. Additionally, I added features and bug fixes to student facing websites. Working at the ECO team taught me a lot about "Big Data" and the approaches we take with it. It was also a much larger and varied team than I had previously experienced. I worked with UI designers, Front-End specific designers, Data Scientists, and Data Engineers. That experience was really valuable, in terms of getting used to real-world teams.

Working for Professor Matthew Butner, I lead a 3-intern team in developing a matching program for study groups. I designed the core algorithm, an alteration of the Stable Roommates problem. I assigned portions of the project and folded them into the real version. The program worked directly with the Canvas API to both make groups and notify students, so it was very easy to run. It was an excellent learning experience for working in a team. I was exposed to alternative perspectives and balanced timelines against additional added features. We were always trying to balance what features we could get in with the next release. The final project was presented to other CS Professors at Computer Education Research at Davis (CERD).

If you're willing to read this, you should really watch the video. I am immensely proud of this project and all the technical work that went into it. Beyond the fact that it won the competition it was made for, it really exemplifies my passion for CS. Smart Traffic has two sides: hardware and software.

On the physical side, there are TI Launchpads. Each TI Launchpad has two sets of pedestrian walking signals, LED traffic lights, and two Force Sensitive Resistors (FSRs). The FSR are used to measure the weight of little toy cars. The TI Launchpads and server communicate back and forth, sharing information. The TI launchpad will take its orders and control the signals.

The software handles the bulk of the logic. The state of each launchpad is stored and used to calculate the next move. The logic prevents various issues, such as incompatible timing, starvation, and deadlock. There are additional configurations beyond the usual traffic lights. For example, the forward and left turn lane of only one side of a street can be turned on at once, while the opposite side pedestrian light is moving.

Several potential modes exist. One, "Priority Mode" allows the traffic light to be overridden, most likely for a first responder. Others take into account things like pedestrian safety, a mode where all four sides of the traffic stop for the pedestrians to cross. Of course, there are modes that take into account the weighted plates and give priority to the configuration that allows the most cars to run.

Credit cards have been giving cash back offers and sign on bonuses for some time. "Churning" is the concept of maximizing one's cash back by collecting cards based on their specific offers. For example, a frequent traveler may prefer a card with a 5% hotel offer, even if that card has a $95 dollar annual fee. On the other hand, someone who does not may prefer a card that gives 6% back on groceries. Unfortunately, the process of finding cards and determining what ones work best for you can be time consuming. Existing comparators, like Nerd Wallet, do not take into account the cards you already own and your budget. Credit Card Comparator does.

Credit Card Comparator is very easy to use. First, estimate your annual expenses in each of the categories. Then, check off the cards you already own from the list. Then, submit your response and wait for a comprehensive list.

email: [email protected]