ML & AI Engineer — Fine-Tuning · Agentic Systems · Edge Deployment · Production LLM Ops
Dallas-Fort Worth, TX · ttimmsinternational@gmail.com · Portfolio
I build production LLM systems from the metal up — from quantized models running on Jetson edge hardware to multi-agent cloud deployments with tool-use, permission gating, and audit trails. Currently focused on MoE fine-tuning, Blackwell-native FP4 quantization (NVFP4), and agentic coding benchmarks (SWE-bench).
Open to ML Engineer roles — DFW or remote.
Security-first open-source coding agent. Hand-rolled async ReAct loop with 4-tier deny-first permission engine, SHA-256 hash-chained audit trail, and 200+ LLM providers via LiteLLM. 4,600+ tests.
- SWE-bench Lite (dev-23 split): 34.8% single-shot · 52.2% oracle best-of-5 — free-tier drivers, $0 API spend; methodology
- 30+ built-in tools with JSON Schema validation, MCP server + client
- Parallel + speculative tool dispatch, cost budget enforcement
- Self-evolution via LLM-guided mutations, multi-language verify gate with retry
- Training data export (openai/chatml/sharegpt), per-step reward annotations for GRPO
End-to-end NVFP4 W4A4 quantization + serving for Zyphra's ZAYA1-8B (80-layer MoE + CCA attention) on a 16 GB RTX 5070 Ti. Rebuilt vLLM from source with SM120 CUTLASS FP4 kernels, wrote the layer-wise activation calibration, and reverse-engineered the NVFP4 global-scale convention.
- 9.5 tok/s single-stream · ~74 tok/s batch-8 (
enforce_eager=True), 6.02 GB checkpoint - n-gram speculative decoding: validated 2.2× speedup on coding-edit prompts, zero training required
- Budget-forced reasoning evals proving checkpoint health: GPQA-Diamond 45.8% → 62.5% with reasoning budget
Autonomous multi-agent personal intelligence system on NVIDIA Jetson Orin Nano Super. 5 LangGraph expert agents, LiteLLM gateway (4 providers + Ollama), 3-tier ONNX intent router. 415 tests. Fully on-device — zero cloud dependencies.
Multi-agent algorithmic trading pipeline with DeepSeek R1 reasoning at every stage. 4-agent pipeline (TA → Chief → Risk → Execution), Kelly Criterion position sizing, Monte Carlo risk simulation, real-time WebSocket market data.
Qwen3.5-4B fine-tuned with ORPO for biblical Q&A. Hybrid RAG (ChromaDB + BM25 + cross-encoder reranking), constitutional AI guardrails, voice pipeline (Whisper + Kokoro TTS), Gradio UI. 183 tests, 34 W&B runs, 5,925 training steps.
- GPU Server Test Suite — GPU fleet validation modeled on NVIDIA DCGM, Prometheus + Grafana dashboards. 188 tests.
- ML Lab — Experiment lifecycle control plane, orchestrates gpu-server-test-suite and llm-wiki. 49 tests.
- LLM Wiki — Git-backed knowledge base, LangGraph ingest/query pipelines, BM25 search. 117 tests.
- ML Experiment Scaffold — GitHub template for reproducible single-GPU ML experiments (config-driven, 3-seed reporting).
- Manufacturing Quality Analytics — SQL + Python ETL for semiconductor quality analysis.
- Tesla Tire Wear ML — Multi-model ensemble for tire wear prediction.
- llama.cpp #22897 — NVFP4 default type mapping + per-tensor scale tensors + MSE-optimal correction (open, awaiting upstream review)
- llama.cpp #22858 — Missing
LLAMA_FTYPE_MOSTLY_NVFP4case fix (closed, replaced by #22897)
📈 Contribution Graph
| Area | Technologies |
|---|---|
| LLMs & Agents | LiteLLM, 200+ providers, Ollama, llama.cpp, multi-agent orchestration, ReAct loops |
| Fine-Tuning | Unsloth, TRL (SFT/DPO/GRPO/ORPO), QLoRA, PEFT, MoE architectures, RLHF/RLAIF |
| Inference | vLLM (custom CUTLASS builds), llama.cpp, n-gram speculative decoding (2.2× validated) |
| Quantization | NVFP4 (Blackwell-native), GGUF, FP8, NF4, GPTQ, AWQ |
| ML Infrastructure | PyTorch, CUDA 12.8, torch.compile, DeepSpeed, lm-eval, W&B, MLflow |
| Systems | Python, TypeScript, Docker, GitHub Actions CI/CD, systemd |
| Edge / Hardware | NVIDIA Jetson Orin Nano Super, RTX 5070 Ti (Blackwell sm_120), 16 GB VRAM optimization |
| Data | PostgreSQL, SQL, pandas, SQLAlchemy, ChromaDB, LanceDB, BM25 |