Some might not see an obvious path from biomedical engineering and math to machine learning, but Shipra Arjun did and used it to make her way to Ontra. As an engineer on the Ontra machine learning team, she is continually experimenting with and learning ways to improve the Ontra user experience. Read on to discover how she and the team keep pace with and apply the latest technologies, tools, and models.
Tell us a little about your background and how you came to Ontra.
I’ve always been interested in engineering and exploring how things work. As an undergrad, I double majored in biomedical engineering and math with a minor in computer science. I spent time in two research labs: a biomedical lab and medical imaging lab. In the biomedical lab, I studied the impacts of low-frequency vibrations on the musculoskeletal system. In the medical imaging lab, I used computer vision and machine learning to segment colorectal cancer tissues to automate the detection of abnormalities seen in scans.
After graduating, I started as a software engineer in the defense industry. Since then, I have worked in the health tech and insurance sectors as well. My projects ranged from designing systems that were used in the defense space to building machine learning models to provide better cost transparency to patients. In each of these projects, I used different types of data to design, build, and scale solutions.
Despite starting as a software engineer, I wanted to get back into machine learning after working on computer vision problems in the research lab; it also seemed like a natural progression. The common thread between biomedical engineering, math, and machine learning is problem solving and understanding the scientific process of forming a hypothesis, designing experiments, and presenting the data.
I found out about Ontra when I was a reference for a prior colleague who was joining Ontra’s machine learning team. Using machine learning in the legal field is imperative because of the complex workflows that lawyers rely upon. It’s exciting to leverage machine learning and its tools at Ontra to automate those workflows because I can see and contribute directly to the contract negotiation process.
In your opinion, what sets Ontra’s machine learning team apart?
The team represents many engineering backgrounds and deep expertise in different domains, whether algorithms, machine learning systems and operations, or math and statistics. With everyone contributing different insights and viewpoints, there is a collaborative approach.
Plus, everyone on the team is constantly experimenting – the field is constantly evolving and there are many significant advancements being made in natural language processing, generative AI, and large language models. It’s all part of the collaboration that’s important for building models, staying atop the latest research, and improving existing models.
What have you found most rewarding about working at Ontra so far?
I‘m involved in projects that challenge me. One effort that comes to mind was improving the transparency of our model’s performance. This transparency not only builds trust and accountability, it also helps drive effective decision-making and engagement from other Ontra teams. Plus it facilitates continuous improvement by allowing us to identify areas for improvement and any biases or concerns.
At Ontra, there’s a culture of continuous learning, and I am encouraged to keep developing as a professional. Though I had worked as a machine learning engineer before joining Ontra, this was my first time working in the natural language processing space. There is so much to learn, and I’m encouraged to explore the field and learn about different techniques and tools.
It’s a great space to be in right now as emerging techniques in natural language processing gain traction that can help tackle more intricate contract automation problems. We can apply machine learning to create — sometimes more sophisticated and sometimes simpler — models to address the many complexities involved in contract negotiations.
What recent machine learning innovation at Ontra most excites you?
More recently, there’s been a greater emphasis on integrating AI into every facet of the Ontra product suite. That’s exciting because we get to work closely with product teams and align on how machine learning can improve the user experience and drive value for our customers. Because of this, we can accelerate and improve the negotiation process in many areas. We also see how users interact with the product, which enables us to iterate on models and features more intentionally so we can deliver more exciting and impactful features for our users.
For example, we can harness machine learning algorithms and natural language processing to extract relevant parts of contracts to not only speed up the negotiation workflow but also potentially help detect patterns to identify potential risks or ambiguities.
What advice would you offer someone interested in a career in machine learning?
As much as it’s important to understand theory — such as how algorithm choices impact a model — machine learning is a practical skill and the best way to learn is by doing.
Don’t get overwhelmed by all that you’ll find online — the math syntax, complex algorithms, data sets, and different programming languages. Start with simple projects and work your way up to complex ones. Find a topic you want to explore and play with the data sets, many of which are open source.
We still do this within Ontra to keep up with the latest advancements and make sure we don’t miss an opportunity to use a new tool or technique. It’s important for us to experiment to see how a model or data set might be applicable, or if any existing Ontra models could benefit.
What’s a fun fact that few know about you?
A couple of years ago, I had a two-week gap between jobs. At the last minute, I decided to go on a solo backpacking trip to the island of Kauai in Hawaii. I explored by foot for much of it, but during the trip, I went on a hiking and kayaking excursion. I kayaked eight miles alone, and then met up with locals as I hiked on the Nepali coast.
I also went snorkeling off a random beach. That was my best snorkeling experience and so serendipitous because I hadn’t planned on it and saw a family of sea turtles below me.
