Hey there, I'm Ashwin Patil.

I am a Data Engineer at Fetch, where I create event-driven, petabyte-scale real-time systems with single-digit microsecond latency for Machine Learning Recommender systems. This allows me to work with tools like AWS, Kubernets, Kafka, Kafka Connect, Flink, Airflow, DBT, and Snowflake, built on the Kappa architecture.

I completed my Master's in Computer Science with a specialization in Natural Language Processing, Information Retrieval, and Deep Learning from the University of Illinois Urbana-Champaign. During my time there, I worked as a Graduate Research Assistant, advised by Prof. ChengXiang Zhai at the Text Information Management and Analysis Group (TIMAN).

Python is my preferred programming language for anything and everything, from developing and training deep learning models and creating a pixel-perfect website for myself to writing basic daily automation scripts. I can also code in Java, Go, C++, Ruby, C, and Javascript.

Currently, the focus of my research lies at the intersection of Natural Language Programming (NLP), Information Retrieval (IR), Scalable Systems, Knowledge Representation and Reasoning, and Deep Learning (DL). Basically, this a messy intersection of buzz words and cool stuff. This is where I operate.

Imagine you’re standing at a chaotic crossroads in a bustling city. There’s a signpost in the middle, and the signs are pointing in every direction with fancy names: “Natural Language Programming (NLP),” “Information Retrieval (IR),” “Scalable Systems,” “Knowledge Representation and Reasoning,” and “Deep Learning (DL).” Each of these places sounds like it could either be the setting of a sci-fi movie or the name of a really complicated board game.

Think of NLP as a translator who speaks every human language and also happens to be a robot. This guy can read, write, and chat like a human but never gets tired or makes grammar mistakes. Cool, right? But it gets messy when he starts learning from the internet because, well, the internet is like a gigantic, noisy cafeteria where everyone’s shouting at the same time.

Next, you’ve got IR, which is like the world’s most diligent librarian. This librarian knows where every book, article, and random post-it note is in the universe of information. When you ask for something, they sprint off to fetch exactly what you need, even if it’s buried under a mountain of cat memes and Wikipedia entries.

Then there’s the land of Scalable Systems. Imagine you’re throwing a party, and suddenly everyone in the city wants to come. Scalable Systems is like having an endless supply of pizza, drinks, and seating that magically appears as your guest list grows. No one goes hungry, and there’s always enough room for more people to join the fun.

Now, Knowledge Representation and Reasoning is like that friend who’s ridiculously good at trivia. This friend not only remembers everything but also connects the dots in ways that blow your mind. They don’t just know that Paris is the capital of France; they can also tell you why it’s historically significant and how it’s connected to other global events. They’re like a living, breathing encyclopedia with a Ph.D. in Everything.

Finally, we have Deep Learning, the brainy superhero of the group. This guy takes in vast amounts of data and learns to recognize patterns, predict outcomes, and even create art. Imagine if Sherlock Holmes and Picasso had a baby who grew up to be a genius—yep, that’s Deep Learning.

What is "Hic Sunt Dracones"?

Hic Sunt Dracones translates from Latin to "Here Be Dragons." Historically, this phrase was used on medieval maps to denote unexplored or perilous areas, warning travelers of potential dangers. I particularly like this because it symbolizes the mystery and adventure inherent in the unknown; while the known brings comfort and certainty, the unknown always makes for better stories.


This blog is self-hosted on AWS and Cloudflare using the open-source Ghost platform.