16-year-old Developer | Python | AI/ML | Trading Bots
Self-taught developer who wrote my first line of Python at age 9. By 16, I have shipped production-grade trading bots, a multi-agent AI framework, and a document intelligence API -- all deployed and running in the real world. I learn best by building things that work.
- Python -- FastAPI, asyncio, pandas, NumPy, scikit-learn, ONNX Runtime
- Trading Systems -- Kalshi API, Kelly criterion, VaR models, shadow-mode validation, kill-switch architecture
- AI / ML -- RAG pipelines, HuggingFace Transformers, FAISS vector search, LLM agent orchestration
- Infrastructure -- Docker, GitHub Actions, Raspberry Pi (headless), Prometheus monitoring, systemd
Currently engineering -- reinforcement-learning position sizing, multi-modal document parsing, Kubernetes deployment for hobby clusters.
Seeking -- mentorship from senior engineers, open-source collaboration, internship opportunities where I can ship real code.
| Category | Tools / Libraries |
|---|---|
| Languages | Python (3.10+), Bash |
| Web / API | FastAPI, Uvicorn, Pydantic, HTTPX, REST, WebSockets |
| Data / ML | pandas, NumPy, scikit-learn, HuggingFace Transformers, ONNX Runtime, FAISS |
| Trading / Finance | Kalshi API, ccxt, Kelly criterion, VaR, backtesting frameworks |
| DevOps / CI | Docker, GitHub Actions, pytest, black, ruff, Prometheus |
| Hardware | Raspberry Pi (Zero 2 W, Pi 4), headless Linux, GPIO |
| Databases | SQLite, PostgreSQL, Redis (basic) |
| Domain | Proficiency | Details |
|---|---|---|
| Document Intelligence | Expert | OCR pipelines, layout parsing, pdfplumber, Tesseract, multi-format extraction |
| LLM Integration | Advanced | RAG with FAISS, OpenAI/Claude API, prompt engineering, agent orchestration |
| Model Serving | Advanced | FastAPI + ONNX Runtime, Dockerised inference, CPU-optimised, sub-second latency |
| Trading Algorithms | Expert | Kelly sizing, VaR 95% limits, kill-switch architecture, shadow-mode validation, backtesting |
| Multi-Agent Systems | Advanced | Agent orchestration, sandbox execution, task decomposition, code generation pipelines |
| Computer Vision | Intermediate | OCR preprocessing, image classification, Pillow, OpenCV basics |
doc-intel-api -- Document Intelligence REST API
FastAPI service that extracts structured data from PDFs and images using OCR and layout analysis. Designed for production deployment on low-cost hardware.
| Aspect | Details |
|---|---|
| Stack | Python, FastAPI, Tesseract, pdfplumber, Pillow, Docker |
| Scale | ~200 requests/min on Pi 4 (4 GB) |
| Performance | Avg 350ms per page, CPU-only |
| Security | Input validation, file-type whitelist, rate limiting |
| Repo | github.com/DavidEscotoDev/doc-intel-api |
Production-ready document extraction pipeline built from scratch. Handles PDFs, scanned images, and multi-page documents with structured JSON output.
Framework -- Multi-Agent Coding Assistant
Lightweight framework for orchestrating LLM-based agents that autonomously generate, review, and test code. Control plane with sandboxed execution environment.
| Aspect | Details |
|---|---|
| Stack | Python, FastAPI, asyncio, Docker sandbox, OpenAI-compatible API |
| Scale | 5 concurrent agents, extensible architecture |
| Performance | Agent turnaround < 2s for standard tasks |
| Safety | Container-isolated execution, static analysis hooks |
| Repo | github.com/DavidEscotoDev/Framework |
Orchestration engine for autonomous code generation. Agents collaborate on complex tasks with human-in-the-loop validation.
kalshi-pi-bot -- Trading Bot for Raspberry Pi
Headless trading bot running 24/7 on a Raspberry Pi, connected to Kalshi prediction-market API. Order placement, position tracking, and risk controls.
| Aspect | Details |
|---|---|
| Stack | Python, asyncio, aiohttp, SQLite, systemd |
| Scale | 24/7 operation on Pi Zero 2 W, < 5% CPU |
| Reliability | Auto-reconnect, exponential backoff, health-check endpoint, Prometheus metrics |
| Repo | github.com/DavidEscotoDev/kalshi-pi-bot |
Always-on trading infrastructure running on $15 hardware. Combines networking, persistent state, and production monitoring.
kalshi_bot -- Advanced Trading Bot with Risk Management
Sophisticated trading system with Kelly-criterion position sizing, Value-at-Risk limits, kill-switches, and shadow-mode validator that simulates trades before committing capital.
| Aspect | Details |
|---|---|
| Stack | Python, asyncio, pandas, NumPy, PostgreSQL, Prometheus |
| Scale | 100+ markets simultaneously, sub-second order execution |
| Risk Controls | Kelly sizing, VaR 95%, max-drawdown guard, audit log, shadow mode |
| Repo | github.com/DavidEscotoDev/kalshi_bot |
Quantitative trading system with institutional-grade risk controls. Shadow mode validates every trade decision before real execution.
PAPYRUS-AI -- LLM-Powered Document Intelligence
Natural-language interface over document collections using open-source LLMs via ONNX Runtime for fully on-device inference.
| Aspect | Details |
|---|---|
| Stack | Python, ONNX Runtime, FastAPI, FAISS, Streamlit |
| Scale | Indexes thousands of documents, ~1s query latency |
| Architecture | RAG pipeline with vector search, reranking, and context synthesis |
| Repo | github.com/DavidEscotoDev/PAPYRUS-AI |
On-device RAG engine. No cloud dependency -- runs entirely locally with privacy-first architecture.
| Period | Role | Highlights |
|---|---|---|
| 2023 -- Present | Independent Developer | Designed, built, and deployed 5+ production projects including trading infrastructure, AI microservices, and agent orchestration systems. Wrote CI/CD pipelines, comprehensive tests, and operational monitoring. |
| 2022 -- 2023 | High School Coding Club Lead | Founded and led weekly Python workshops. Mentored 15+ peers on Git, GitHub, and web development. Built club infrastructure with Flask and automated deployment. |
| Recognition | Details |
|---|---|
| GitHub Arctic Code Vault Contributor | Code preserved in the 2025 GitHub Archive snapshot. |
| First Place -- School Hackathon (2024) | Built real-time sensor dashboard on Raspberry Pi with live data visualisation. |
| Open Source Contributor | Active contributions to Python tooling and AI/ML open-source projects. 100+ commits across public repositories. |
In progress / Planned:
- AWS Certified Cloud Practitioner -- Studying, exam Q4 2026
- Python Institute -- PCEP (Certified Entry-Level Python Programmer) -- Scheduled
- DeepLearning.AI -- TensorFlow Developer Certificate -- Curriculum in progress
- freeCodeCamp -- Scientific Computing with Python -- 400+ hours of structured computer science curriculum completed
Each certification represents a milestone in a structured engineering education path I designed for myself.
- LeetCode -- leetcode.com/DavidEscotoDev -- 150+ problems solved, weekly contests
- Codeforces -- codeforces.com/profile/DavidEscotoDev -- Rating 1300+, division 3
learning:
- "Reinforcement learning for trading strategy optimisation"
- "Kubernetes and container orchestration for ML workloads"
- "Advanced MLOps: MLflow, DVC, model registry, A/B testing"
- "CUDA programming and GPU-accelerated inference"
building:
- "RL-based position sizing engine for kalshi_bot v3"
- "Multi-modal document parser combining OCR + vision transformer"
- "Self-hosted agent orchestration platform with web UI"
exploring:
- "Contributing to PyTorch and HuggingFace open-source"
- "Building a small-scale GPU cluster for distributed inference"
- "Technical writing on trading systems and AI infrastructure"
open_to:
- "Mentorship from senior engineers in ML/systems/trading"
- "Part-time internship or apprenticeship -- will ship real code"
- "Collaborative open-source projects in AI infrastructure"
"Built from scratch since age 9. Always shipping."