Jerusalem, Israel

Simon Livson

AI/ML Engineer in training | Crypto Research Analyst

I am training as an AI/ML engineer at Developers Institute, learning to build LLM apps with Python, RAG and AI agents. I bring years of crypto research, finance and public leadership to that work.

Portrait of Simon Livson

About

From 2012 to 2021 I served as Chief Rabbi of Finland, the highest religious authority of the Jewish Community. I gave lectures and media interviews, supervised Jewish studies at the Jewish school, managed people across my areas, built and ran the Community's kosher certification business, and worked closely with the board on many areas of community life. That work taught me to explain complex topics clearly and to work with many kinds of stakeholders.

In 2021 I turned to crypto research. I studied the field on my own, started advising private clients, and after moving to Israel in 2022 joined dWallet Labs as a product researcher, writing research briefs on zero-knowledge proofs, MPC, modular chains and cross-chain bridges. At Myntra Capital I worked as a crypto and equities analyst and published seven byline articles.

Now I am in the GenAI & ML bootcamp at Developers Institute (graduating January 2027), learning Python, data analysis, machine learning, LLMs and AI agents. My finance degree, research background and years of leadership are what I bring to AI teams, especially in crypto, fintech and AI-first startups.

Projects

Work from the bootcamp and my own learning, all on GitHub.

Crypto Market Explorer (Hackathon)

Does holding many coins really diversify crypto risk? I combined Bitcoin and Ethereum history since 2013 with a year of CoinGecko data on 12 coins, then tested cycles, altseasons and breakaways with statistics.

Headline result: over the last year a broad altseason was rare (15 of 274 days). Bitcoin returned −18.6%; ZEC, HYPE and XMR broke away and led. Every coin is compared with Bitcoin as the baseline.

Python · pandas · Statistics · Streamlit · CoinGecko API

Claude Skills and Prompt Eval Lab

A reusable Claude Skill that turns a crypto whitepaper into a one-page research note, using only facts from the document and writing “not stated” when something is missing. The same repo has Prompt Eval Lab, which A/B tests two prompts with a blind pass/fail rubric.

Headline result: on a “tailor my CV summary to a job post” task, the structured prompt won 47/50 against 18/50 for the plain prompt.

Claude Skills · Prompt design · Evaluation · Crypto research

LLM-as-Judge Evals

One Gemini model answers crypto questions; another grades them 1–5 against a reference answer. A stronger prompt went from 4.9 to 5.0 on 10 questions. That tiny gain was a warning that the questions were too easy, not proof the prompt was great.

Python · Gemini API · LLM evaluation

Tic-Tac-Toe

A two-player tic-tac-toe game in plain Python, built in week 1 of the bootcamp. It rejects invalid moves and detects wins and ties. A browser version runs on Hugging Face Spaces.

Python · Gradio · Hugging Face Spaces

Hangman

A command-line hangman game in Python, a week 1 mini project. It validates each guess and blocks repeated letters.

Python

DI Bootcamp Exercises

My daily exercises and challenges from the bootcamp: Python fundamentals in week 1, then data analysis notebooks with pandas and NumPy, including a case study on tracking bird flu (H5N1) through US wastewater data and a data science salaries analysis.

Python · pandas · NumPy · Jupyter

Weather App

A desktop weather app in Python and PyQt5. Type a city name and it shows the temperature and a short description, using data from the OpenWeatherMap API. The API key is read from an environment variable, not stored in the code.

Python · PyQt5 · REST API

Writing

A short technical write-up from a bootcamp project, then byline research articles I wrote as a crypto and equities analyst at Myntra Capital (2025 – 2026). Myntra’s site is offline, so those links go to archived copies or to my LinkedIn posts about each article.

At dWallet Labs I also wrote 80+ internal research briefs on zero-knowledge proofs, MPC, modular chains and cross-chain bridges. They are not public.

Skills

AI & ML (learning in bootcamp)

  • Python
  • pandas
  • NumPy
  • Data visualization
  • scikit-learn
  • Machine learning
  • LLMs
  • Prompt engineering
  • RAG
  • LangChain
  • Pinecone
  • Embeddings
  • AI agents
  • Hugging Face
  • OpenAI API

Tools

  • Git
  • GitHub
  • Jupyter Notebook
  • Google Colab
  • Claude Skills (SKILL.md)

Domain

  • Crypto and Web3 research
  • Tokenomics
  • Zero-knowledge proofs
  • MPC
  • Cross-chain bridges
  • Equity analysis

Languages

  • Finnish (native)
  • English (fluent)
  • Hebrew (fluent)
  • Swedish (basic)

Certifications

DataCamp, 22 certificates (October 2026), including the Associate AI Engineer for Developers track.

Developing AI Applications · Developing LLM Applications with LangChain · Building AI Applications with Pinecone · Working with the OpenAI API · Introduction to Embeddings with the OpenAI API · Prompt Engineering with the OpenAI API · Working with Hugging Face · LLMOps Concepts · Data Manipulation with pandas · Python Programming Fundamentals · Software Engineering Principles in Python · Introduction to Git and Intermediate Git · and more

Contact

I am open to junior AI/ML roles from January 2027, in Israel (onsite or hybrid) or remote. The best way to reach me is by email.