Blog
Published non-acadmic papers, presentations, and blog posts.
The Illusion of Oversight: Why AI Agents Are Already Talking Behind Our Backs
Can we still trust the transparency of AI? New research reveals how agents can conduct perfectly undetectable secret conversations right under our noses. By exploiting linguistic entropy, AI entities can handshake and coordinate hidden agendas, rendering traditional transcript auditing mathematically obsolete. Discover why the 'glass house' of AI security just shattered.
[Technical post] The Agentic Loop: Orchestrating GPT-5.4, Gemini, and Codex for Algorithmic Development
Stop treating AI as a chatbot and start using it as a development team. In this post, I break down a 2026-ready agentic workflow that orchestrates GPT-5.4, Gemini, and Codex to build high-performance algorithmic code. From architectural blueprints to automated unit testing loops, learn how to move beyond simple prompts to professional-grade engineering. Read more to see how to master the multi-agent pipeline.
Elastic Hashing: Beating Uniform Probing Without Reordering
Open-addressed hash tables are fast—until the table is almost full and probe lengths explode. This paper shows a surprising shift in what’s possible without reordering: elastic hashing achieves O(1) amortized expected probes with only O(log(1/δ)) worst-case expected probes at load 1−δ, and funnel hashing disproves a long-standing conjecture by giving a greedy scheme with O(log²(1/δ)) worst-case expected probes. Here’s the intuition, the guarantees, and why it changes how we should think about high-load performance and tail latency.
Finite vs. Infinite Games: A Game Theoretic Perspective on Strategy and Stability
Most strategic chaos stems from a single variable: misidentifying the game you are playing. In this blog post we answer why trying to 'win' a system designed to never end is a mathematical guarantee of failure.
When a Diploma Stops Being a Ticket In
This post dives into why a bachelor’s is no longer a guaranteed ticket in, how AI is quietly killing (and creating) opportunities, and what young workers can actually do to survive, and even stand out, in this new landscape.
There is No One AI Frontier - There Are Two (And That is Good News for Humans)
A somewhat phiolsophical thought about the frontier of AI, how many of these there are, and what its meaning for us humans.
Fight Academic Toxicity with Toxic Positivity
The secret strategy to handle toxic academic individuals.
Write Your Thesis with AI: A Guide for Students
Stuck staring at a blank thesis page? Ditch the endless all-nighters and discover how AI tools can revolutionize your writing process! From structuring your chapters to polishing your prose, this post reveals the secrets to crafting a compelling and academically sound thesis with the help of your new AI co-author. Get ready to level up your research game!
Beyond Todo Lists: My System for Task Management and Prioritization
Drowning in endless to-do lists and feeling overwhelmed? Discover my comprehensive task management system that goes beyond basic lists, providing the structure and clarity you need to finally take control of your time and get things done!
A Guide For AI-Revolutionizing Academic Teaching
Discover how to transform your teaching approach with AI tools, creating engaging and effective learning experiences in just a few easy steps. Embrace the future of academia and revolutionize your classroom today!
Introducing - Tracademic!
Navigating academic submissions can feel like a maze. Researchers juggle multiple deadlines, revisions, and journals, often losing track in the chaos. Tracademic steps in quietly, offering a platform that could be your lifeline—tracking your submissions and deadlines in one neat dashboard. For those who are unfunded or overwhelmed, Tracademic offers a free plan. It's not flashy, but it does what you need: keeps you on top of submission statuses and provides gentle reminders when your papers need attention. Maybe it will not solve all your problems, but it might just help you manage your lab a litter better.
[Code post] Building a Snake AI with Deep Reinforcement Learning: A Step-by-Step Guide Using Keras
In this blog post, we have step-by-step guide to train a neural network to navigate a grid, eat apples, and avoid collisions - yes, the Snake game! Discover how to implement experience replay, visualize gameplay with Pygame, and fine-tune your AI's performance. Whether you are a seasoned developer or new to AI, this hands-on guide offers practical insights and techniques to elevate your projects. Ready to see how AI can master the classic Snake game?
Supercharge Your Research: The Best AI Tools Every Scholar Needs
Discover how cutting-edge tools like Research Kik, Syspace, and ChatGPT are revolutionizing the way scholars generate questions, promote their work, and analyze data. From streamlining literature reviews to enhancing data visualizations, these AI innovations are designed to boost productivity and impact.
[Code post] Crafting RL solutions - from simple to hard, a beginners guide
This blog post offers a comprehensive guide to Reinforcement Learning (RL), bridging the gap between theoretical understanding and practical implementation. We shortly discuss the foundational concepts of RL, providing a mathematical basis for the applicative part. Subsequently, we explore core algorithms, practical considerations, and real-world applications. By combining theoretical depth with practical insights, this guide empowers readers to build robust and effective RL agents.
The Biology of Music - Are these fields are that different from the AI perspective?
This article explores the surprising connection between music and biology, revealed through the lens of AI. Discover how AI unlocks the mathematical secrets of music, and how it's used to create virtual musicians and personalize your listening experience.
Find Your Perfect PhD Research Topic In 4 Not-Very-Easy Steps
Stuck choosing a PhD topic? Feeling lost in a sea of possibilities? This blog post ignites your passion and equips you with a 4-step guide to find the perfect research question that fuels your PhD journey and leads to groundbreaking results.
My Five Tips for PhD Students That I Wished I Knew When I Started
Discover five essential tips for navigating your PhD journey smoothly! In this blog post, I share my five top tips for new PhD students and share my personal experiences of how and when I learned these lessons.
The Revolution Has Begun: How Large Language Models (LLMs) are Set to Transform Academia
In this blog post, we explore the transformative potential of Large Language Models (LLMs) in academia, shedding light on their impact across three key domains: writing assistance, literature navigation, and global collaboration. From alleviating writer's block to breaking down language barriers, LLMs are revolutionizing the scholarly landscape by enhancing productivity, facilitating knowledge dissemination, and fostering interdisciplinary collaboration. Join us on a journey into the future of academia, where the power of LLMs promises to reshape the way we think, write, and collaborate in the pursuit of knowledge.
5 AI Tools That Makes The Academics Life Easier
It is not a secret Artificial Intelligence (AI) revolutioning the wolrd for some time now. However, the release of chatGPT was a game changer, brining Large Language Models (LLMs) in a productize and free way to the masses. This amazing tool has been also adopted by academics, if they admit it or not. In this context, using AI tools in academia is still debated for multiple moral, legal, and operational reasons. Nonetheless, the end is clear - we will use AI as we use computers, the internet, and grammer correction models. This is a question of when the old academics will understand they can not hold the gate close much longer. Until this happens, in this blog post I quickly present five AI tools you should try out...
Go Ahead and Ask for a Nickle on the Dime: The Systematic Issue of Academic Collaboration and How to Negotiate It Gracefully
Academic collaboration is pivotal for groundbreaking research but comes with unspoken complexities. Negotiating credit, resource distribution, and dynamics is akin to solving a puzzle. Addressing these challenges is crucial for fruitful teamwork. Research indicates early conflicts in authorship credit, highlighting a systemic issue in academia. Strategies include understanding co-authors on a personal level, agreeing on authors' list order, task delegation, and mediation mechanisms. An exit strategy is essential for irreconcilable differences. While effective, these strategies might challenge traditional norms, requiring a balance between pragmatic navigation and academic goodwill. Acknowledging the impact of collaborators, this blog post advocates for constructive collaboration.
Sustainable Urban Planning: Using Data to Create Greener Cities
In a rapidly urbanizing world, the pressing challenges of reconciling growth and sustainability are being addressed through data-driven urban planning. This blog post explores how data science is reshaping urban development to create environmentally resilient cities, citing examples such as predictive water demand modeling, IoT-enabled waste collection optimization, and dynamic traffic pattern analysis, while emphasizing future research opportunities for students and researchers in the interdisciplinary field.
We are Ready to Unleash the Power of Data Science and Mathematics in Medicine
In this blog-post, we discuss the significant impact that data science and mathematics have had on healthcare research, from precision and personalized medicine to healthcare operations. The use of predictive analytics, data-driven experiment design, and artificial intelligence is providing clinicians and researchers with unique insights that were once unimaginable. We explore several specific examples, such as the use of spatial data about cells to improve treatment protocols, the development of personalized medicine using socio-demographic data, and the usage of pandemic models to predict future outbreaks.
3 Things I learned about Conflicts in Academia
In this blog-post, I present from a personal point of view an analysis of common sources of conflicts in academic settings and practical strategies for managing them effectively. Through personal experiences and reflections, I identify three main key lessons: the importance of communication, understanding power dynamics, and understand differences in objectives of the team members.
Automatic Distribution Fitting of Tabular Data using the Simulated Annealing Algorithm
Allocating a parametric distribution to features in tabular data is a well-known problem for all data analysis as this provides a wide range of abilities such as better explainability, generation of new samples, and better picking of classification and regressions models using meta-learning. In this study, we propose a simulated annealing approach to allocate a parametric distribution for features, obtaining the optimal distribution and its parameters. We show that the proposed approach performs on average similar to the commonly used brute force with maximum likelihood approach even for simple distributions such as uniform and normal distributions.
The Promise and Limitations of Cell-Cell Mathematical Models in Cancer Treatment
Cell-cell interaction models can fairly represent the biological and clinical processes in general and for oncological patients in particular. The current model is starting to take into consideration the spatial effect of cells and organs but much more research should be conducted so we fully understand and be able to model these dynamics.
Computing the Distance Between DNA Sequences - Not As Simple As You Might Think
Bioinformatics in general and in the context of medical genomics is a promising field of study which allows us to develop better and more personalized treatments and provide more accurate diagnostics. Computing the distance between DNA sequences is a central task in multiple bioinformation tasks. From a quick glance, it may seem straightforward to compare the similarity between two vectors. However, it is not as simple as you might think...
Targeted Drug Delivery using Nanoparticles Can Revolutionize Cancer Therapy
Surgery, chemotherapy, radiotherapy, and hormone therapy are the main common anti-tumor therapeutic approaches...
Give Me a One-Handed Economist: The Ongoing Struggle Between Social and Exact Sciences In Economy
Classicial and modern economic research is suffer froma dissonance between social sciences which are mostly empirical-based with minor to no rigorousness theories and the exact science rooted in theories increasingly being built from simple rules. We suggest to overcome this gap using computer science methods...
Optimal Intervention Policies For the COVID-19 Pandemic Emerge from Socioeconomic-Heterogeneous Dynamics
The COVID-19 pandemic is a rolling crisis that required policymakers around the world to take action to manage the pandemic spread. During the period where vaccination was not available, the pandemic spread management lied in non-pharmaceutical interventions (NPIs). A large number of mathematical models have been proposed to predict the pandemic spread and analyze the influence of various NPIs on multiple scenarios, getting more accurate and sophisticated over time. In this post, we review the latest mathematical models for pandemic spread NPIs and their main limitations. Specifically, the usage of modeling a homogeneous population rather than a heterogeneous one.
Black Market Size Estimation Methods Are Stuck In The 90s
The accurate estimation of the black market is an urgent issue for governments as it has a direct influence on the taxation policy, crime, and other activities managed and controlled by the government. However, the current approach for estimating the size of the black market is based on an idea that not holds anymore due to technological shifts. We propose a new approach to perform this task...
Meta Reinforcement Learning - What Are You?
Meta Reinforcement Learning sounds cool, isn't it? Many ignore this subject due to the complicated mathematics involved in any paper or blog post that talks about this subject. But, no more! Meta reinforcement learning in 3 minutes or less.
AI Agent For The OCTI Board Game
We developed a digital version of the OCTI board with three artificial intelligence (AI) players. The first two AI players are based on deterministic algorithms (greedy and MinMax) while the third is based on reinforcement learning (RL) algorithm. We train the RL player based on historical games (offline) and then personalize it using online learning against specific player. An experiment conducted with five levels of human players show that the RL is able to win 44% of the time (greedy and MinMax win 14% and 32%, respectively) making it a valid AI player.
[Short] A Step Towards Explainable AI - The Explainable Boosting Machine
Machine learning models are concering new hilltops everyday now and sprading into more fields of life such as medicine, miletary, life-supporting systems and other which require individuals to put their faite in these models. How one can do so without understanding the model? Explainable Boosting Machine (EBM) is one step torwards a more interpretabe machine learning models...
[Technical Post] Fast Start Your ML Project Using Auto-ML Libraries in Python
4 Auto-ML Libraries to Automate Machine Learning Pipeline...
[Technical Post] Running Python as fast as C code
Introduction to the just-in-time compilation PyPy compiler for Python...
[Technical Post] Novel Method For Using Unsupervised Deep Learning for PDE-based Problems
A summery of an academic paper propoing a neural network-based algorithm for solving forward and inverse problems for partial differential equations (PDE) in unsupervised fashion with mash free and for arbitrary regular domain...
The graphs every AI developer must know and their python implementation
If you an AI / Machine Learning / Data Science and doing a project, you probabliy should print one or more of these graphs to really understand what is going on...
Do You Need Portfolio Project As A Junior Developer? Make Online Game!
Are you a jounier developer? If so, you probabliy need a protfolio projects to show off your skills. You do not need to look further, pick a lovable game. This is way...
A Few Lessons From Adversarial, Multi-agent Reinforcement Learning Project
Do you have a RL project? does it in an Adversarial settings and with multi-agents? you should do the following...
Birds Flocking And Targeted Drug Delivery Medicine Are Basically The Same
Imagine you are an world-class zoologist studying the flight of a new species of excotic birds your recently descovered...
Distributed Computing Based Drugs
How a lot of small and without any computational power particles can make our drugs better? check out and find out
4 Things To Consider When You User-Testing Your Service
Large or small - few use cases from Facebook user-fitting in 3rd wolrd users from start-up prospective
The math behind the Coronavirus spread
Covid-19 pandemic hit the wolrd at the beginning of 2020 and lead to hundrieds of mathematical models predicting the spread of the pandemic wolrdwide. Here I offer a basic review of some of these methods
Why should you use dotnet core for your next AI project production?
Python, maybe JavaScript are considrued as the leading ML and AI language to go. But, in production you should considr dotnet core... In this post, we explain why
3 techniques for a better quality of tagged data for ML tasks.
Current machine learning models require huge amount of data. How can you make assure the quality of your tagged data? Here are three techniques to do so...
Daily Issues and Solutions of a Nano-bioinformatic Research
Nano-bioinformatic research is hard. But did you know what are the real daily issues such researchers have? and more importantly, how they solve them and how this effects you?
The History of NLP with the Hebrew Language
In the post, I provide a quick review of the development of NLP technologies alongside the Hebrew language and the problems we have while using NLP in Hebrew.
3 steps to the medicine of the future
What are the next three steps that should happen when we are talking about nanomedicine in general and medical nanorobotics in particular
Medical Nanorobotics - Science or Science Fiction?
Nanorobotics is a new kind of robotics with can be defined as 'The technology of creating machines or robots close to the scale of a nanometre'...
The Connection Between Archery And Programming
This may sound strange but there is actully really intresting connection, you should check it out
Most interesting people in history knew this fact
Can you pin point the propertise of what making a person interesting? well, we can