I was born in Whitefish, Montana and attended Lewis & Clark College in Portland Oregon, where I earned my degree in Mathematics. Professionally, I'm a data scientist and software developer who loves to try new languages, APIs and Python packages. My current projects include text localization and recognition in real-world images, an app called Emojiface, and building
I was born in Whitefish, Montana and attended Lewis & Clark College in Portland Oregon, where I earned my degree in Mathematics. Professionally, I'm a data scientist and software developer who loves to try new languages, APIs and Python packages. My current projects include text localization and recognition in real-world images, an app called Emojiface, and building a better Twitter Recommendation system using Spark and Scala. I also love writing music, skiing, and hiking in my free time.
I have experience using:
- Python (numpy, pandas, scipy, scikit-learn, flask, etc.)
Feel free to reach out to me on LinkedIn if you have any questions or would like to get to know me better!
Software Developer @ Summary:
- Core team member
- Developed facial feature eraser in Python
- Integrated external C++ libraries like dlib and OpenCV, translated code to C++ for integration
- Provided business strategy and marketing recommendations to improve UX
As a contract software developer at EmojiFace, I applied my knowledge of image processing to build a facial feature eraser to eliminate friction between the user and application. Abstracting images as matrices, I utilized linear algebra and popular image processing libraries to develop a novel method to localize facial features (eyes, eyebrows, mouth) and blend them into the background of the face. This process eliminated prior anomalies caused by improper lighting and the shape subject's resting face. I am currently awaiting further work as we reassess priorities for the next iteration of the product. From September 2015 to Present (4 months) Greater New York City AreaData Scientist @ Metis is a technical academy that offers a 12-week Data Science program covering topics including:
- Software development in Python
- Statistical analysis of large data sets
- Data mining, cleaning and storage
- Machine Learning, Regression Analysis
- Relational and non-relational databases (PostgreSQL and MongoDB)
- Hadoop, MapReduce, Hive
Metis focuses on project-driven learning. Over the course of the program, I built projects that use Python to import, analyze and visualize my findings and present them as a finished product to my peers as I would do so to a data science team. My projects, which can be reviewed in full on GitHub, are listed below:
1. MTA Subway Data - Identified New York’s busiest subway stations with publically-available data, geocoding and data visualization.
2. Predicting Oscar Winners - Mined IMDB data to predict Academy Awards by user and critic reviews, box office earnings, and other relevant features.
3. Optical Character Recognition - Used popular classifier algorithms (KNN, Naive Bayes, Random Forests) to categorize real-world images of English characters obtained from Google Maps. Current Top Ten entry on Kaggle.
4. Twitter Recommendation System - Used clustering methods and NLP to analyze a user's twitter presence and provide content-based recommendations for who to follow with the goal of improving user experience.
Final project: text localization and recognition in real-world images. I designed a program that ingensts an image and localizes text candiates using a slew of image-processing algorithms and heuristics. The application also utilizes NLP to provide better guesses in the case that the computer makes an error in isolating characters. From June 2015 to Present (7 months) Greater New York City AreaMarketing Specialist @ Summary:
- Optimized HTML for relevance and content on client websites
- Managed Social Media profiles for Clients
- Wrote Python and VBA tools for data management and entry
- Used A/B tests to generate actionable business decisions
- Analyzed data from Google Analytics to provide recommendations and analyze performance
As a marketing specialist at Adpearance, I was responsible for a number of tasks in the field of Search Engine Optimization. These tasks ranged from writing and implementing new copy and HTML for websites, optimizing existing HTML elements and conducting keyword research to flesh out plans for clients in industries ranging from heavy machinery to legal services. I was also required to manage clients' social media presence, audit websites for possible improvements, and conduct A/B tests on client websites to suggest improvements. Beyond the normal requirements of the job role, I was also asked to improve and expand our reporting by using more approachable statistics and finding ways to quantify SEO success in reports.
Apart from my job description, I additionally took on the responsibility for developing tools to be used on my team, ranging from simple jargon-checkers to check for brand consistency in text to more in-depth Python scripts to inject data into client webpages, automating some of the more tedious parts of dealing with a CMS by hand. These tools cut anywhere from 5 minutes of manually checking documents for misbranding of a client's name or ill-formatted jargon, to close to an hour of manually copying and pasting data into online forms. From February 2015 to June 2015 (5 months) Portland, Oregon AreaResearch Intern @ Research Topic: Empirically Evaluating and Quantifying the Effects of Inspections and Testing on Security Vulnerabilities
- Mined Gerrit repositories using Java app
- Constructed, filtered and queried a MySQL Database
- Analyzed and Explored data using R
- Produced peer-reviewed research, sponsored at FSE Conference 2014
Over the course of nine weeks, I collected and analyzed data on security vulnerabilities and the corresponding project-experience level of those who introduced and found them. By mining Gerrit repositories with a Java applet, my research team acquired large amounts of code-review data which we filtered (by way of SQL queries and manual inspection) down to a data set which we could statistically analyze.
I was responsible for designing statistical tests to answer the questions my team had developed throughout the course of the project. For example, one of our greatest areas of interest was how the experience of a given developer (measured in months that they had contributed to the project) correlated to the likelihood of introducing security bug. The goal of this project, which my team acheived, was to produce work that could be published in journals and presented at conferences globally. The recent publishing of our paper at FSE 2014 in Hong Kong was a great acheivement for me personally, as well as for my team. From May 2013 to July 2013 (3 months) Tuscaloosa, Alabama Area
Bachelor of Science (B.Sc.), Mathematics @ Lewis and Clark College From 2011 to 2014 University of Montana From 2010 to 2011 Derek Janni is skilled in: Statistics, Mathematics, Java, C++, Mathematical Modeling, Numerical Analysis, Qualitative Research, Python, Probability, Applied Mathematics, Computer Science, R, Quantitative Research, Data Science, Data Analysis
Looking for a different
Get an email address for anyone on LinkedIn with the ContactOut Chrome extension