Agentic-AI Engineer • OSS Maintainer • Sr GenAI & ML Engineer @ Thomson Reuters Special Services
- 📍 Based in Washington DC — shipping multilingual RAG & governance tooling for regulated industries.
- 🔬 Inventor on 9 patents covering dynamic-location sampling & sensor-fusion analytics.
- 🎓 B.S. Mathematics (UMass Amherst) • M.S. Computer Science (Wentworth Institute).
Theme | Current Focus |
---|---|
DevEx | 💚 Self-healing CI/CD (Git rerere , union merge drivers, auto-rebase action). |
Supply Chain | 🔐 semantic-release → Cosign key-less signatures + SBOM diff bot. |
Observability | 📊 Real-time DORA dashboards & token/latency/cost collectors for LLM apps. |
🕸️ Agent Frameworks & Orchestration
Repo | Stars | What it does |
---|---|---|
agentic-dev-orchestrator | WSJF backlog → AutoGen / CrewAI multi-agent CI. | |
agent-mesh-federated-runtime | P2P runtime (10 k agents, BFT). | |
async-toolformer-orchestrator | Parallel tool-calling for GPT-4o 5× API. | |
reflexion-agent-boilerplate | Plug-in self-reflection / self-improvement. | |
autogen-code-review-bot | Two-agent PR reviewer loop. | |
agent-skeptic-bench | Truthfulness adversarial bench. | |
agentic-startup-studio | Automate ideation → MVP → pivot loop. | |
agentic-startup-studio-boilerplate | Cookiecutter skeleton (CrewAI + FastAPI + React). | |
agentauth (private) | – | Delegated credentials & Okta "Auth for GenAI". |
agent-orchestrated-etl | LangChain ⇔ Airflow hybrid ETL. | |
crewai-email-triage | Smart inbox classifier / replier. | |
slack-kb-agent | Slack bot answers from team docs. | |
agentic-redteam-radar | Cloud-Security-Alliance-inspired agent scanner. |
🔐 Secure & Private AI
Repo | Stars | Highlight |
---|---|---|
homomorphic-llm-proxy | GPU-accelerated FHE gateway. | |
secure-mpc-transformer-infer | Non-interactive MPC BERT (< 1 min). | |
privacy-preserving-agent-finetuner | DP LoRA fine-tuning. | |
rlhf-audit-trail | Immutable RLHF provenance. | |
dp-flash-attention | Flash-Attention 3 + Rényi DP. | |
tokenwatch (private) | – | Proxy logs tokens / cost / jailbreaks. |
fair-credit-scorer-bias-mitigation | Fair lending model toolkit. | |
hipaa-compliance-summarizer | Redacts PHI → compliance notes. | |
redactai (private) | – | PII/PHA redaction pipeline. |
pqc-migration-audit | Auto-migrates crypto to PQC. |
📐 Evaluation & Governance
Repo | Stars | Highlight |
---|---|---|
agi-eval-sandbox | One-click DeepEval + HELM. | |
agent-skeptic-bench | Epistemic humility probes. | |
deception-redteam-bench | Detect emergent manipulation. | |
causal-eval-bench | Causal reasoning offline probes. | |
llm-cost-tracker | OTEL token/latency/cost collector. | |
lm-watermark-lab | Evaluate watermark robustness. | |
modelcard-as-code-generator | Auto-generates model cards. | |
self-healing-pipeline-guard | Autorepair CI pipelines. | |
cot-safepath-filter | Sanitises chain-of-thought. |
⚛️ Quantum & Analog ML Toolkits
Repo | Stars | Highlight |
---|---|---|
quantum-devops-ci | CI/CD for hybrid QML. | |
pqc-iot-retrofit-scanner | PQC upgrade planner for IoT. | |
quantum-agent-scheduler | QUBO-based agent planner. | |
quantum-mlops-workbench | DevOps for quantum pipelines. | |
quantum-inspired-task-planner | Q-annealing / classical fallback. | |
analog-pde-solver-sim | Analog computing for PDEs. | |
photonic-mlir-bridge | MLIR → silicon-photonic compiler. | |
photonic-nn-foundry | Full stack for photonic AI HW. | |
dp-federated-lora-lab | DP federated LoRA. | |
open-moe-trainer-lab | End-to-end sparse MoE training. | |
dynamic-moe-router-kit | Adaptive expert routing. | |
pno-physics-bench | Probabilistic Neural Operators. |
🛰️ Edge & Neuromorphic
Repo | Stars | Highlight |
---|---|---|
liquid-edge-lln-kit | Liquid NN on Cortex-M/ESP32. | |
tiny-llm-edge-profiler | 2-bit/4-bit LLM edge bench. | |
fast-vlm-ondevice-kit | < 250 ms VLM on iPhone. | |
spikeformer-neuromorphic-kit | Spiking transformer toolkit. | |
mobile-multi-mod-llm | 35 MB vision-text LLM. | |
iot-edge-graph-anomaly | GNN-augmented anomaly detect. | |
wasm-shim-for-torch | Run PyTorch in browser (WASI-NN). | |
edge-tpu-v5-benchmark-suite | First open benchmark for TPU v5e. |
🧬 Scientific Discovery & Simulation
Repo | Stars | Highlight |
---|---|---|
self-driving-materials-orchestrator | 10× faster materials lab. | |
neural-operator-foundation-lab | Training + bench for neural ops. | |
protein-diffusion-design-lab | Diffusion for protein design. | |
tokamak-rl-control-suite | RL for plasma shape. | |
pg-neo-graph-rl | Dynamic GNN + FedRL control. | |
dynamic-graph-fed-rl-lab | Time-evolving graph RL. | |
genRF-circuit-diffuser | Cycle-consistent RF design. | |
surrogate-gradient-optim-lab | Learned gradient surrogates. | |
neoRL-industrial-gym | Real-world offline RL bench. | |
analog-pde-solver-sim | Analog PDE solver simulator. |
⚡ Developer Tools & Utilities
Repo | Stars | What it does |
---|---|---|
openapi-doc-generator | Auto-writes OpenAPI schemas & human docs from code. | |
docker-optimizer-agent | Generates minimal, secure Dockerfiles + SBOM. | |
crewai-email-triage | CrewAI inbox classifier / auto-replier. | |
testgen-copilot | Suggests unit tests & vuln patches from diff. | |
async-toolformer-orchestrator | Parallel tool-calling runtime for GPT-4o. | |
agentauth (private) | – | Delegated credentials & Okta "Auth for GenAI". |
hf-eco2ai-plugin | Logs CO₂, kWh & grid mix for each HF epoch. | |
nimify-anything | Wrap any ONNX/TensorRT artefact into an NVIDIA NIM micro-service. | |
Code-Mesh | Repo-wide embeddings & semantic code queries. |
📊 Data Analysis & Modeling
Repo | Stars | Highlight |
---|---|---|
llm-tab-cleaner | LLM-powered data-quality & cleaning pipeline. | |
surrogate-gradient-optim-lab | Learned gradient surrogates for black-box optim. | |
dynamic-graph-fed-rl-lab | Fed RL for time-evolving graph control. | |
neoRL-industrial-gym | Offline RL benchmarks for industrial control. | |
iot-anomaly-detector-timeseries | Multivariate sensor anomaly detection via LSTM-AE. |
🖼️ Vision, Media & Multimodal
Repo | Stars | Highlight |
---|---|---|
vid-diffusion-benchmark-suite | Unified latency / quality harness for VDMs. | |
multimodal-contract-extractor | VLM that parses scanned legal PDFs. | |
fast-vlm-ondevice-kit | <250 ms multimodal inference on iPhone. | |
liquid-ai-vision-kit | Liquid Neural Nets for micro-drones & AVs. | |
nerf-edge-kit | On-device NeRF for spatial computing headsets. |
📦 Supply-Chain & DevSecOps
Repo | Stars | Highlight |
---|---|---|
self-healing-pipeline-guard | Auto-fixes red CI pipelines. | |
provenance-graph-sbom-linker | End-to-end supply-chain attestation. | |
modelcard-as-code-generator | Turns eval logs → executable Model Cards. | |
cot-safepath-filter | Sanitises chain-of-thought before it leaks. |
🧑⚕️ Synthetic Data, Fairness & Governance
Repo | Stars | Highlight |
---|---|---|
synthetic-data-guardian | Validated synthetic data + watermarking. | |
fair-credit-scorer-bias-mitigation | Debiases credit-scoring models. | |
biasaudit-pro (private) | – | Annex VII-ready bias auditing SaaS. |
🚀 Hardware & Performance
Repo | Stars | Highlight |
---|---|---|
gaudi3-scale-starter | PyTorch Lightning boilerplate for Intel Gaudi 3. | |
edge-tpu-v5-benchmark-suite | First open harness for Google TPU v5e edge cards. | |
dp-flash-attention | FlashAttn-3 kernels + Rényi DP noise, zero-overhead. | |
wasm-shim-for-torch | Run PyTorch 2.4 in WebAssembly (no WebGPU). |
Where | Link |
---|---|
[email protected] | |
linkedin.com/in/daniel-schmidt-574482b8 | |
📝 Blog | (auto-generated via openapi-doc-generator , coming soon) |
"Build systems that heal, models that explain, and agents that behave."