All software in this portfolio was designed and built by Oliver Ellison d/b/a Reliable AI Network, Inc. ("RAIN").
A single place to explore production-grade agentic AI work. Each project includes a short summary, use cases, tech stack, and screenshots. Built to be skimmed by engineering leaders, execs, and recruiters.
Contact: oliveraellison@gmail.com • LinkedIn
Live Portfolio Grid: https://aurelius-in.github.io/agentic-portfolio/
Source of truth: docs/portfolio.json
Repo: https://github.com/aurelius-in/AI-Health-Navigator-for-Patients
Summary
Healthcare navigation with intelligent triage, provider matching, and personalized recommendations using multi-agent reasoning with memory (episodic, semantic, long-term) and cross-agent collaboration.
Use cases
- Symptom intake to recommended care path
- Provider matching and appointment guidance
- Benefits and coverage checks with safety and compliance context
Tech stack
- Backend: FastAPI, PyTorch, PostgreSQL, Redis, ChromaDB
- Frontend: React, TypeScript
- Infra: Docker, Kubernetes, Prometheus, Grafana; HIPAA-aware controls
Screenshots
Repo: https://github.com/aurelius-in/AutoOps-Sentinel
Summary
Agentic AIOps that ingests metrics and logs, detects anomalies, forecasts incidents, and executes safe auto-remediation with a modern React dashboard. Explains actions, answers questions, and quantifies cost avoided.
Use cases
- Proactive incident prevention and SLO burn detection
- Automated runbooks with policy guardrails
- Exec-ready narratives and value attribution
Tech stack
- Backend: FastAPI, Python
- Frontend: React, TypeScript (MUI)
- ML: Z-score, IsolationForest, MAD, Prophet
- Ops: Docker, GitHub Actions CI/CD, RBAC, Slack and Teams integration
Demo
Repo: https://github.com/aurelius-in/Claims-Triage-AI
Summary
Agent-driven decision support for claims classification, risk scoring, routing, and compliance with explainable AI and human-in-the-loop oversight. React dashboards deliver analytics, audit trails, and visual model explanations.
Use cases
- Insurance claims intake and triage
- Healthcare prior authorization and medical necessity review
- Finance and legal case routing with policy-as-code
Tech stack
- Backend: FastAPI, PostgreSQL, Redis, Docker
- AI/ML: LLM and ML hybrid, SHAP explainability, retrieval, OPA for policies
- Frontend: React, TypeScript; Prometheus and Grafana observability
Demo
Repo: https://github.com/aurelius-in/MindTrace
Summary
Enterprise wellness platform with five agents (Companion, Recommendation, Risk Detection, Analytics, Privacy) for organizational health, burnout detection, and resilience with a privacy-first design.
Use cases
- Employee wellness companion and mood journaling
- HR analytics on anonymized, privacy-safe signals
- Burnout risk prediction and intervention guidance
Tech stack
- Backend: FastAPI, PostgreSQL, Redis, ChromaDB or Pinecone
- Frontend: React, TypeScript
- Integrations: Slack, Teams, HRIS, Outlook
- Ops: Kubernetes and Helm, Prometheus and Grafana, OPA; differential privacy
Demo
Repo: https://github.com/aurelius-in/high-volume-recruiter-ai-agent
Summary
End-to-end recruiter flow (outreach → consent → screening → scheduling → ATS sync) with a live ops UI, real-time KPIs, and a signed, replayable audit trail. Runs offline for controlled demos or in real mode via adapters, with capacity/SLA analytics and policy-checked messaging.
Use cases
- High-volume roles (retail, logistics, CX): scale outreach, screening, scheduling
- Staffing and RPO teams: live ops console with agent assist and policy guardrails
- Compliance and audit: signed ledger and replay for verification and incident review
Tech stack
- Backend: Python 3.11, FastAPI, Uvicorn; SSE endpoints (
/events/stream,/chat/stream); signed audit chain - Frontend: React + Vite + Material UI, Recharts, i18n (EN, AR, ZH), dark mode; Ask panel with streaming
- Infra/QA: Docker; Postgres-ready seams; pytest; structured logs and PII redaction
Demo
Repo: https://github.com/aurelius-in/AI-Governance
Summary
Enterprise AI governance with an LLM proxy, policies-as-code, safety guardrails, automated compliance checks, and real-time cost monitoring. Built with production-grade MLOps and full observability.
Use cases
- Policy enforcement and redaction on LLM traffic
- Cost controls and vendor routing
- Trust, safety, and auditability for AI systems
Tech stack
- Backend: FastAPI, PostgreSQL, Redis
- Governance: OPA, guardrails, automated checks
- Ops: Docker and Kubernetes, Prometheus and Grafana; React frontend
Screenshots
Repo: https://github.com/aurelius-in/Smart-Document-Bot
Summary
Multi-agent document analysis for regulatory compliance and business intelligence. Ingest, classify, extract entities, assess risk, and monitor compliance with audit trails and live agent traces.
Use cases
- Regulatory document intake and QA
- Risk flagging and policy checks
- Interactive analytics and evidence export
Tech stack
- Backend: FastAPI, PostgreSQL, vector search
- Frontend: React, TypeScript
- Observability: OpenTelemetry; containerized deployment
Screenshots
Repo: https://github.com/aurelius-in/MuseAgent
Summary
A modular multi-agent music intelligence platform that ingests tracks, extracts spectral features, generates embeddings, tags mood, genre, and instrumentation, delivers explainable recommendations with visualizations, exports branded PDF reports, and includes optional enrichment and AI-powered music generation.
Screenshots
Repo: https://github.com/aurelius-in/adops-flightdeck-multi-agent
Summary
Production-grade multi-agent AdOps Flightdeck that orchestrates AI agents for planning, creative generation and safety, pacing, anomaly response, attribution, and auditable reporting on AWS.
Screenshots
Repo: https://github.com/aurelius-in/EdgeSight-QA
Summary
Modular, containerized edge computer-vision QA: multi-camera ingest to on-device inference to SCADA write-backs, with operator UI, metrics, and offline-safe logging.
Screenshots
Repo: https://github.com/aurelius-in/MyRiskAgent
Summary
Agentic risk analysis that ingests claims, filings, and OSINT to compute provider and company risk scores with peer-normalized outlier detection, SHAP explainability, and audit-ready evidence.
Tech stack
FastAPI (Python 3.11), SQLModel/SQLAlchemy, DuckDB, Postgres/pgvector, Redis, React + TypeScript + Vite + MUI + TanStack Query + ECharts, OPA (Rego), Prometheus, Grafana, Docker Compose, Alembic; LangGraph-style orchestration.
Screenshot
Repo: https://github.com/aurelius-in/perception-lab
Summary
Interactive environment for exploring perception scenarios, running detection and fusion pipelines, monitoring performance metrics, and exporting evaluation reports.
Demo
Repo: https://github.com/aurelius-in/ACE-Lab
Summary
Real-time WebGPU-first creative lab that blends agentic shader design, on-device image generation, and microservice video generation using AnimateDiff for motion and RIFE for interpolation, with generative fill, style transfer, and policy-safe WebM export via OPA in WASM.
Screenshots
Repo: https://github.com/aurelius-in/RainRef
Summary
An end-to-end AI support “Ref”: normalize tickets/chats/emails into RefEvents, draft grounded answers with citations, propose safe actions via declarative playbooks, run OPA policy gates before execution, write signed receipts, and emit Product Signals (bugs, frictions, requests, pricing, churn) to Jira/Linear/RainScout.
Use cases
- AI Support Copilot (answers with sources)
- Known-issue autopatcher (safe actions + receipts)
- Churn risk sweeps and objection handling
- Weekly evidence packs for CS/Product
Tech stack
- Backend: FastAPI (Python 3.11), Postgres + pgvector, Azure Blob (Azurite dev), LangGraph flows, OPA/Rego policies, Langfuse, OpenTelemetry
- Frontend: React + TypeScript + Vite
- Ops: Docker Compose (dev/prod profiles), migrations & seed, smoke tests, Make targets
Demo
| Project | Primary Domain | Core Agents | Signals and Models | Governance and Observability | Frontend |
|---|---|---|---|---|---|
| Health Navigator | Healthcare navigation | Planner, Triage, Memory, Provider-Match | LLM with retrieval, clinical triage heuristics | HIPAA awareness, RBAC, audit | React/TS |
| AutoOps Sentinel | AIOps | Detector, Forecaster, Remediator, Explainer | Z-score, IsolationForest, MAD, Prophet | Runbooks, RBAC, CI/CD, Slack and Teams | React/MUI |
| Claims Triage | Insurance and healthcare | Classifier, Risk, Router, Compliance, Support | XGBoost+SHAP, retrieval, OPA | Audit, PII redaction, policies-as-code | React/TS |
| MindTrace | Workforce wellness | Companion, Recommender, Risk, Analytics, Privacy | Sentiment and risk models, DP analytics | HIPAA, GDPR, SOC2, OPA | React/TS |
| High-Volume Recruiter | Recruiting ops | Outreach, Consent, Qualifier, Scheduler, ATS Sync | Policy checks, SSE streaming, multilingual i18n | Signed audit ledger, replay, SLA heatmaps | React/Vite/MUI |
| AI Governance | Governance platform | Proxy, Policy, Safety, Cost | Guardrails, OPA, usage telemetry | Cost and safety enforcement | React |
| aiDa | Document intelligence | Orchestrator plus task agents | NER, QA, compare, translate | Audit, OpenTelemetry | React/TS |
| MuseAgent | Music intelligence | Ingest, Feature Extractor, Embedder, Tagger, Generator | MFCC, mel, chroma, CLAP/OpenL3, RAG | Deterministic demo mode, Dockerized, eval reports | React/TS |
| AdOps Flightdeck | Advertising ops | Planner, Creative QA, Pacing, Anomaly, Attribution, Reporter | LangGraphJS, Bedrock models, anomaly detectors | AWS-native, OpenTelemetry, auditable reporting | React/TS |
| EdgeSight QA | Industrial CV | Ingest, Infer, QA, SCADA Writer, Monitor | ONNX/TensorRT, rules and thresholds | Offline logging, Prometheus, Grafana | React/Vite/MUI |
| MyRiskAgent | Risk and compliance | Ingest, Score, Explain, Policy Gate, Report | SHAP, peer normalization, OSINT retrieval | OPA, evidence bundles, immutable logs | React/TS |
| Perception Lab | Perception R&D | Detector, Fusion, Monitor, Evaluator | CV pipelines, COCO metrics, IDF1 | Audit-ready exports, Docker | React/TS |
| ACE Lab | Real-time creative | Shader, Generator, Video, Policy Gate | WebGPU/WGSL, AnimateDiff, RIFE | OPA to WASM, export safety | React/Vite |
| Support & CS Ops | Intake, Grounder, Action Planner, Policy Gate, Signalizer | BM25+pgvector retrieval, OPA receipts | React/Vite UI |
Each project has or targets a skimmable one-pager in docs/:
- Problem and business value
- Agent architecture and orchestration
- Data flow and storage
- Safety, privacy, and governance
- Deployability and ops notes
- Demo script and wow moments
File naming: docs/<project-id>-onepager.md
Examples: docs/health-navigator-onepager.md, docs/autoops-sentinel-onepager.md, docs/myriskagent-onepager.md
MIT