The Startup Internship — DeepSig Inc.
This blog serves the purpose of highlighting my journey in securing an internship and then working for a startup in Summer 2018, while I was studying for my Masters at Virginia Tech.
Scoring an internship while you are in grad school is a very competitive and a mind-numbing process. It involves hundreds of applications and an almost equal amount of rejections! The key thing is not to lose hope and keep applying to places where you think you’d fit well. Additionally, the career fair at Virginia Tech is not very conducive for students who are looking for jobs in Machine Learning, Data Sciences, and Deep Learning. In fact, most of the companies are looking for Software Development/ Engineering roles and thus I had to do the majority of my applications online.
Sometime in March, I came across DeepSig Inc through one of the Virginia Tech’s listservs. The company’s web page has a huge banner stating their work as reinventing wireless with deep learning. I was curious to know more about it and hence I applied for an internship position at the place. I promptly received a reply from Ben Hilburn, the company’s Director of Engineering with whom I had a phone interview where we discussed my work experience and mainly, I asked questions about the company. The next part was a technical presentation of my work to the entire company (of 5 people) over a video call!!!! A couple of emails later, they told me that I had secured the internship. So, I packed up my bags in May and headed to Arlington, Virginia (3 metro stops away from DC) to start my internship. I joined a team of 5 people who were situated in a co-working space in the posh(like suits,boots, and dresses posh!) neighborhood of Clarendon. The startup was co-founded by a Virginia Tech Professor Tim O’Shea and an experienced startup founder Jim Shea (who I assure you are not related to each other).
Following are some of the key differences or points that are different when you’re working at a startup as compared to a major established company-
Complete look at the product life cycle —
It was a learning experience to work on a software product and watch it develop from a simple idea to a revenue-generating product! My work was initially focused on OmniSIG sensor which is a product used for RF spectrum sensing using Artificial Intelligence. I was responsible for integrating the product with gnuradio, an open-source software-defined radio toolkit. Hence, I got to see how an algorithm working (in parts and pieces) on a Jupyter notebook is converted to a full-fledged working, license-able and revenue generating software!
New Day, New Challenge!
Tasks that were offloaded to me were extremely topic agnostic. Basically, as and when the requirements changed or we got a new deadline for our products, the priorities changed for everyone. I got a chance to research deep learning literature to find a solution for a couple of days, I also worked on integration of the existing product with open source libs, at one point I was also working on a producing relevant results(which I failed to do so) for an article that we were writing for Nvidia! In addition to that, the work was fairly inter-disciplinary. I got exposed to various topics from broadband communications, electronics fundamentals, deep learning as well as computer architecture. It made the software engineering job extremely tasteful and added a lot of flavor to my experience there.
Workload Flexibility—
Since the company was in a discovery phase with respect to its product, customers as well as the work that was needed to be done, the workload was never fixed!
A task that I took up was modeling properties of a wireless channel using deep learning and also creating a synchronized version of transmitter/receiver pair over a noisy channel. My work helped us collect data from radio channels which we wished to model for our product, OmniPHY. This was more of research role, wherein fundamentals in communication were useful to support the software engineering work needed to complete the work. Initially, this was a no-deadline task since it was more research-oriented, I spent a lot of days reading and implementing various algorithms that I felt might help solve the problem and hence pace of the work was in manner-of-speaking ‘slow’. In fact, I worked from home (Blacksburg, since I also had to complete some grad school formalities) for almost a week. But, once we had a more defined product along with a deadline to complete a well-defined piece of it, the pace of work quickened as the work was more defined! This flexibility in the pace helped me keep a good work-life balance.
Extremely Informal Culture —
Our office had a couple of beer taps, which had a variety of new brews every week or two, so my team also enhanced my palate for a beer with their expert opinions on different brews! We also visited a few local breweries near Clarendon which totally changed my opinion about how snobbish one could be about their favorite brew!
The first few days I felt extremely dwarfed by the achievements of my peers as they were a bunch of extremely talented, creative and knowledgeable folks around me. This informal culture really helped me form great inter-personal relationships! Rather than saying I met new colleagues, I would say I gained a new set of friends.
The other fun stuff!
Everyone on the team was an experienced open source contributor/maintainer of major projects such as gnuradio and volk. This enabled me to learn more about the open source community and start contributing to the same.
The best part that I had a chance to work on was a project with NASA (the geek in me was excited AF!). Sadly I was not able to travel to their base station to see my work in action (ITAR + International student= 😞) but working on something that personally excites me was an extremely satisfying and thrilling experience!
All in all, I would say that it was an extremely exciting, educational and a fun summer internship, I would like to thank the entire team for making it so! It was awesome to work with people who are equally excited about the ML/AI revolution that is up in the works and who plan to make complete use of their knowledge coupled with awesome tech that we have to solve some really complex problems. It was supremely satisfying to be a part of that solution and contributing towards it!
I would definitely recommend people with background in signal processing and communications, having interest in working on applying deep learning concepts to solve real-world problems to definitely apply for a job here or get in touch with the team here!
