Ramtin Ardeshirifar - Personal Website

Profile

Adept AI and ML researcher and software engineer with a MSc in Artificial Intelligence and Adaptive Systems from University of Sussex.
Developed WiFi Warden, an Android app with over 45 million downloads globally, showcasing strong technical and project management skills, coupled with comprehensive handling of financial, legal, and ethical aspects.
Skilled in Android development using both Java and Kotlin, with a deep understanding of the Android SDK, architecture components, and various third-party libraries.
Having experience with transformers, LLMs, and generative models.
Proficient in Python, TensorFlow, and PyTorch.
Proficient in Python, data collection, data cleaning, and data analysis, with a focus on identifying anomalies, patterns, and correlations.
Successful development and execution of machine learning projects, including an award-winning WiFi analysis tool and an AI mobile game, reflecting capability in applied AI and ML.
Passionate about building smooth, reliable, and delightful mobile experiences.
Effective in a collaborative environment, demonstrated through comprehensive project and team management in prior roles, coupled with excellent communication skills.

Education

University of Sussex, Brighton, UK
MSc, Artificial Intelligence and Adaptive Systems (Distinction) 09/2022 – 09/2023

Achieved distinction in key areas such as Algorithmic Data Science, Mathematics and Computational Method, Machine Learning, Web Applications and Services, NLP, Artificial Life, Adaptive Systems, Image Processing.
Thesis focuses on differentiating human-authored and machine-generated texts using advanced ML and NLP.

Azad University, Tehran, IR
BSc, Industrial Engineering (First-Class Honors) 10/2017 – 09/2021

Thesis on predicting backorders in the supply chain via ML earned the highest grade of 20/20.
Specialized in courses like Data Analytics, Management Information Systems, Decision Analysis, and Operations Research.

Experience

Software Engineer/ ML Engineer (Nuconcepts) 01/2017 – 09/2022

As a Software Engineer/ ML Engineer, I was responsible for the end-to-end development of both the 'WiFi Warden' and 'Rock Paper Scissors with AI' applications, including conceptualization, design, implementation, and refinement.
For 'Rock Paper Scissors with AI', I integrated user interaction data to enhance gameplay and user experience, utilizing Tensorflow Lite for gesture recognition, and continuously improved the AI model based on user feedback and gameplay data.
In developing 'WiFi Warden', my focus was on analyzing and enhancing app performance through the integration of ML algorithms, resulting in improved functionality, efficiency, and heightened user engagement.
Technologies used: Python, Tensorflow, Numpy, Pandas, scikit-learn, Java, Kotlin, C++, AndroidX Libraries, Material Components, Coroutines, Retrofit, Google Play services, Android shell, AppCompat, Junit, Espresso.

Web Developer (Freelance) 01/2014 – 09/2022

Crafted a professional website for Mag Electric company in 2020 and maintained it until 2022. Additionally, initiated a tech news portal (xti.ir) in 2014, where I authored articles on technology trends.

Highlighted Projects

Detecting Machine Generated Text: The findings of this research reveal several intriguing disparities between human and AI text generation. I demonstrated that these differences could be successfully utilized by classifiers to distinguish between human and AI-generated text.
Watermarking on BART Model using CNN Dataset (Conditional Text Generation): In this project my objective was to explore the domain of watermarking algorithms as they apply to Conditional Text Generation (CTG) with a focus on the BART model and the CNN dataset.
Automated Design of Symmetric Autoencoders Using Genetic-Algorithms: Presents a novel methodology for optimizing symmetric autoencoder architectures using genetic algorithms.
StockGPT: Leverages GPT-4 to extract insights from news and disseminate through a dedicated Telegram channel.
Anomaly Detection on Stock-Market using Autoencoders: Implements an anomaly detection system for stock market data using TensorFlow.
Attention Based LSTM Network for Predicting Times Series Data: Employs an LSTM model integrated with an attention mechanism for forecasting currency pair rates.
Credit Card Fraud Detection: Investigates various techniques for detecting fraudulent transactions, including Catboost, undersampling, oversampling, deep learning and anomaly detection.
Predicting YouTube Dislikes using Machine Learning: Applied VADER sentiment analysis and bag-of-words techniques to predict dislikes on YouTube. Details available in a published article.
Word Clouds with Sentiment Analysis of Recent Tweets: Extracted keywords from recent tweets of top news agencies, visualized with a word cloud, and conducted sentiment analysis for in-depth insights.
Traffic forecasting: In this project, I tried to forecast traffic in 6 crossroads with Multi-step dense, GRU and LSTM models. The predictions include 1 step and 3 steps ahead using one and all features.

Profile

Education

Experience

Highlighted Projects

Certifications