37-Month GenAI-First Career Transformation
Business Ops → GenAI-First Data Analyst & AI Engineer → GenAI Data Engineer + AI Systems → ML Engineer + Local LLM Specialist → Agentic AI Engineer → Senior LLM Engineer
While most candidates learn traditional tools only, this journey integrates GenAI/LLM engineering systematically from Week 1—building production AI systems with LLM SDKs (Anthropic primary, Gemini/OpenAI fallback), RAG pipelines, Multimodal AI, FastMCP server, and Streamlit by Month 1. Every project includes evaluation-driven development (DeepEval + pytest, RAGAS, SelfCheckGPT) and Docker containerization. Progressing through Vector DBs, local LLMs, to MCP + A2A + multi-agent orchestration with named patterns (orchestrator-workers, sequential, evaluator-optimizer) by Month 34. Two flagship trading systems anchor the portfolio: Attention-Flow Catalyst (read-only small-cap swing research) and Crucible (autonomous intraday execution — backtest → paper → live, local-first AI, started first).
1099 Reconciliation ETL Pipeline: Live system at Daybright Financial
Status: Active (Months 1-5)
Focus: Python, SQL + LLM SDKs (Anthropic primary) + RAG + FastMCP + Multimodal AI + Streamlit + DeepEval + Docker
Goal: AI Engineer who knows Analytics
Timeline: Months 6-15
Focus: AWS + BigQuery + Vector DBs + RAG Infrastructure + Docker & K8s
Timeline: Months 16-29
Focus: ML + Ollama + Fine-Tuning (PEFT) + NVIDIA DLI
🔥 Crucible: Phase 2 — NautilusTrader migration + autonomous paper-trading agent crew (local Qwen3/Ollama)
Timeline: Months 30-34
Focus: MCP + A2A + LangGraph + Multi-Agent Patterns (orchestrator-workers, sequential, evaluator-optimizer)
🔥 Crucible: Phase 3 — autonomous live micro-sizing on Alpaca + Schwab/TOS with multi-agent oversight
Timeline: Months 35-37
Focus: LLMOps Evaluation + Production AI
Each project introduces new skills that build on the previous • 8 projects, 2 flagships (AFC: research rigor • Crucible: autonomous execution)
Production system automating retirement plan distribution reconciliation between enterprise financial systems.
💰 $15K+ annual savings • ⚡ 95% time reduction • 📊 Zero errors since deployment
Python • pandas • openpyxl • pytest • GitHub Actions CI
Foundation: ETL + Testing + CI/CDAI-Powered PII-Safe Data Intelligence: Natural language analytics for retirement plan operations with PII protection and AI guardrails.
🤖 LLM SDK (provider-agnostic) + PandasAI + Pydantic structured outputs + PII guardrails
+ LLM SDK + Streamlit + Pydantic + PII HandlingAI-Powered HR Policy Chatbot: RAG chatbot answering policy questions with cited sources. Auto-escalates to HR when uncertain.
🧠 Embeddings + ChromaDB + Semantic Search + RAG Pipeline + Ticket Escalation
+ Embeddings + ChromaDB + RAG + Semantic SearchAI-Powered Distribution Form Validator: Multimodal AI reads retirement plan forms (handwritten checkboxes, signatures), validates, and routes.
👁️ Gemini Vision SDK + Multimodal AI + Business Rule Validation + Email Automation
+ Multimodal AI (Vision LLM) + Document ProcessingAI-Powered Workflow Demand Analysis: 8+ months of OnBase enterprise data enabling data-driven staffing decisions.
🤖 LLM SDK + PandasAI + Plotly + Real production data + Privacy guardrails
🎯 8+ months enterprise data • 📊 Distribution vs Loan segmentation • 💬 AI chat interface
+ Enterprise Real Data + Advanced AnalyticsAI-Powered Streaming Subscription Advisor: Optimizes household streaming spend through AI rotation planning, content search, and savings forecasting.
📺 Watchmode/TMDB APIs + httpx async + AI Rotation Planner + Cost Analytics
+ External APIs + Consumer UX + Optimization EngineAI-Powered Predictive Trigger Analysis: Defensible research system for small-cap stocks with statistical rigor. Evolves through all 5 career stages.
🔬 SEC Form 4 + Wikipedia + News + DuckDB Lakehouse + Walk-Forward Validation
📈 5 trigger types • 🎯 Statistical methodology • 🏗️ Complete LLM career demonstration
+ Statistical Methodology + DuckDB + Async + Multi-SourceStrategy-agnostic backtest → paper → live platform: an AI research analyst proposes strategy improvements that are proved by deterministic backtests behind a sealed out-of-sample vault. Strategies are plugins (IT-1 ORB + VWAP Reclaim ship first). Distinct from AFC: liquid intraday execution vs. illiquid small-cap swing research.
🔥 Own backtest harness → NautilusTrader + Local Qwen3/Ollama + LangGraph agents + Alpaca + Schwab/TOS + Sealed OOS Vault
⚙️ Phase 1 in Stage 1 • 🤖 Agentic paper/live phases in Stages 3–4 • 🦙 Local-first AI (no API fee, data stays local)
+ Event-Driven Engine + Local LLM + Multi-Agent Execution + Broker APIs