Current Projects

---

Oh __ Where Art Thou? - Personal AI Guru [Art Project]

Objective: Generative Language Model to serve as an AI Guru.

“[T]he guru’s job is to show the inquirer in some effective way that they are already what they are looking for.” - Alan Watts

While I am still writing the code, I don’t want to share too much about the technical aspect just yet.

Topics Addressed:

• Data Scraping
• Data Cleaning
• nltk
• Text Generation NN Models

---

Semantic Similarity Exploration

Researching methodologies to get a better metric of Semantic Similarity. Process is still on-going.

Details to come soon.

---

Portfolio

---

MAFAT Radar Detection Challenge

Participated in the 2020 Radar Detection Challenge by The Israeli Ministry of Defense “Directorate of Defense Research & Development” (DDR&D)
Competition Details

The participants’ goal is to classify segments of radar tracks of humans or animals using the I/Q signal matrix as an input. The task at hand is a binary classification task; the tracked objects are either humans or animals. The data is real-world data, gathered from diverse geographical locations, different times, sensors, and qualities (high- and low-signal to noise ratio—SNR).

Working together with three friends from Y-Data, we got to the respective 24th place.

Topics Addressed:

• Data Augmentation
• Convolutional Neural Networks
• Temporal Convolutional Networks
• Spectogram and Scalogram Processing
• AWS EC2/ECR/EFS
• WAND&B
• Streaming Pipeline

---

Industry Project - Author Name Disambiguation

This a truncated and redacted version of the original project. Without the data or the company-provided code, none of the notebooks will be directly usable, but extracting functions and pipelines are viable.

For the industry project of the Y-Data course, I partnered up with Academix and helped them deal “Author Name Disambiguation”. Fundamentally, given a group of academix papers, we wanted to figure out how many unique authors existed, and to assign each paper to their respective author.

To watch the final presentation

Topics addressed:

• Data Scraping
• Data Cleaning
• Supervised Learning Methods (Logistic Regression, DBScan)
• Variational Auto-Encoders
• Word2Vec feature embeddings
• AWS EC2

Results: We improved the baseline F1 Score from 0.9 to 0.937

---

Page template forked from evanca