Reorder the page
Write one line about who you are. A model reorders this copy of my home page and writes a short intro for you. The facts come from the site data.
10 tries a dayThe original order
Simon Rendon Arango
Engineer. Creates software. Uses AI cleverly. Occasionally predicts a football match.
Software and AI engineer in London. AI research at Tessl.
About
I hold an MSc in Computing (Software Engineering) from Imperial College London and a BSc in Systems and Computing Engineering from Universidad de los Andes. Since 2023 I have built LLM systems that run in production. At Glamper I built an LLM travel agent, and bookings grew by more than 50%. At Untap I built a pipeline that extracts KPIs from financial reports with 90%+ accuracy. It cut client data ingestion time by more than 80%. I also built a RAG system and an MCP server over portfolio data there. Since June 2026 I have been on the research team at Tessl. We are architecting the software factory, an AI-native platform where coding agents design, build and verify production software. I lead model evaluation, and I shape context evaluation and code verification.
On the side I build LFM, a football model that is in progress. I also made EvalLens, an open-source tool that checks structured LLM outputs. I am from Medellín, Colombia. I moved to London in 2023 for my MSc.
Experience
TesslAI Research
- Core member of Tessl's research team, architecting the software factory: an AI-native platform where coding agents design, build and verify production software
- Lead model evaluation, benchmarking frontier LLMs to select the optimal model for every agentic task
Stack Python, TypeScript, Agents, MCP, LLMs, Evals
UntapSoftware Engineer
- Built an LLM pipeline for Untap's private-equity platform that extracts KPIs, targets and cashflows from financial reports with 90%+ accuracy. It cut client data ingestion time by more than 80%
- Built a RAG system over BigQuery and Gemini, and an MCP server that gives LLMs access to internal tools
Stack React, JavaScript, Java, Vertex, Gemini, GCP
GlamperSoftware Engineer
- Designed and built a serverless backend on AWS Lambda and API Gateway for bookings, payments and users
- Built an LLM travel agent that plans full trips from the current inventory. Bookings grew by more than 50%
Stack Vue, JavaScript, Nuxt, Python, AWS
BancolombiaIntern Software Engineer
- Built Jira and AWS integrations and a React KPI dashboard for the infrastructure team (Nequi). Ticket resolution time fell by 35%
- Automated cloud migration workflows during the bank’s move from on-premise to AWS
Stack Python, JavaScript, React, AWS
Work
EvalLens
An open-source tool that checks structured LLM outputs against expected schemas and values, row by row, and names the reason for every failure.
Large Football Model
A football model I am building. It will give match probabilities during a game, find similar players and matches, and simulate matches. It is not public yet. The page has a waitlist.
The Constellation
Each star is a degree, a job or a project from my CV. 4 lenses group the stars by topic.
Intus
An iPhone and iPad app for habits and a journal that I am building. Each morning and evening a model writes a short card from your entries. The text goes to the model only for that call and is not stored.
Investor intelligence system
A RAG system and an MCP server that let analysts ask questions about portfolio data in plain English.
Financial data extraction
An LLM pipeline that extracts KPIs, cashflows, targets and named entities from financial documents. I fine-tuned the model and wrote the prompts. Accuracy is above 90%.
Performance regression detection in HPC
My MSc thesis. A CI job for Nektar++ that flags performance regressions from hardware counters with a variational autoencoder. It cut triage time from days to under 2 hours.
Predicting Formula 1 results
My BSc thesis. A model that predicts Formula 1 race winners. About 93% accuracy on race winners, on 70+ seasons of Formula 1 results.
Education
Universidad de los AndesBachelor of Science in Systems and Computing Engineering
Courses Data Structures, Algorithm design and analysis, Software Engineering, Mobile and Web Development, Software Architecture
Imperial College LondonMaster of Science in Computing (Software Engineering)
Courses Reinforcement Learning, Deep Learning, Software Engineering, Machine Learning, Natural Language Processing
Constellation
Each star is a degree, a job or a project from my CV. 4 lenses group the stars by topic.
- EngineeringDegrees and jobs in order
- ML systemsProjects with models
- Side projectsBuilt in my own time
- ResearchDegrees and theses
Skills
CoreUsed the most.
- Agents
- Evals
- Python
- RAG
- TypeScript
StrongKnown well.
- AWS
- CI/CD
- Data visualisation
- Deep learning
- Docker
- Express
- Framer Motion
- GCP
- GitLab
- Java
- JavaScript
- LLMs
- MCP
- Next.js
- Node.js
- NumPy
- Pandas
- React
- Reinforcement learning
- Scikit-learn
- shadcn/ui
- Tailwind CSS
- TensorFlow
WorkingUsed less.
- C++
- Design
- Git
- GitHub
- HPC
- Jupyter
- Nuxt.js
- PyTorch
- Vue
Notes
Intus
An iPhone and iPad app I am building for habits and a journal.
The constellation
Why I made a hand-placed map of my CV, and how its 4 lenses work.
EvalLens
A small open-source tool that checks structured LLM outputs row by row.
Who am I?
A first note. Who I am and what this site is.
Ask
A chat that answers questions about my work from the text on this site. If the answer is not there, it says so.
Ask on the home pageContact
For a question or a collaboration, email me or use the form.
simon@rendonarango.comThe sections use the same data as the home page. The model only picks the order, the folded sections, 1 link and the short texts.