Deployment registry
Every engagement, tracked in the open
Three origins: industry work at Rakuten Symphony, independent builds, and the 24-artifact pipeline of my FDE intensive (IIT Roorkee PGC + FDE Academy). Queued rows are slotted builds from the curriculum — they flip to deployed as I ship them. Academy builds are driven by simulated client briefs and RFPs, and are labelled as such.
| ID | Engagement | Stack | Origin | Status |
|---|---|---|---|---|
| ENG-000★ | RAN performance test automation → Production KPI-testing automation for live radio access network setups at Rakuten Symphony — 70% less manual effort, 50% faster suites. | PythonSeleniumJenkins | Industry | deployed |
| ENG-001★ | Slicematic — AI operations copilot for a pizza outlet → Order management, live tracking, and owner-facing AI insights that turn daily order data into margin-growth decisions. | Next.jsAI insightsVercel | Independent | deployed |
| ENG-002★ | This site's copilot → A retrieval-grounded assistant trained on my deployments and case studies — with a designed failure mode when the model is unreachable. | Next.jsRetrievalGemini API | Independent | in-progress |
| ENG-003 | System dependency map Enterprise dependency graph across 5 interconnected services with feedback loops and failure cascades. | Systems thinkingCausal loopsMiro | FDE Academy · M1 | queued |
| ENG-004 | Business-to-technical spec Mock discovery interview converted into a prioritized backlog and technical specification. | SPIN discoveryINVESTRICE | FDE Academy · M1 | queued |
| ENG-005 | Multi-source data ingestion pipeline 3 APIs + database + CSV unified with data-quality gates via Great Expectations. | PostgreSQLAirflowGreat Expectations | FDE Academy · M2 | queued |
| ENG-006 | Operational dashboard prototype Real-time KPI dashboard with live data refresh for operational decision-making. | StreamlitRetool | FDE Academy · M2 | queued |
| ENG-007 | Production prompt library 20+ structured business prompts with evaluation benchmarks via Promptfoo. | PromptfooDSPyClaude API | FDE Academy · M2 | queued |
| ENG-008 | End-to-end ML pipeline + API Train, evaluate, and serve a model behind FastAPI with MLflow experiment tracking. | Scikit-learnFastAPIMLflow | FDE Academy · M3 | queued |
| ENG-009★ | RAG knowledge system Enterprise document Q&A over a vector database with grounded LLM generation. | ChromaLangChainEmbeddings | FDE Academy · M3 | queued |
| ENG-010 | Consulting toolkit Discovery report, PoV→PoC→MVP strategy, AI-readiness assessment, and client proposal. | ConsultingAI adoption | FDE Academy · M3 | queued |
| ENG-011 | Fine-tuned SLM QLoRA fine-tune of a small language model, served with vLLM, benchmarked against GPT-4o. | QLoRAvLLMHugging Face | FDE Academy · M4 | queued |
| ENG-012 | LLM gateway Multi-model routing with fallback logic, cost tracking, and latency monitoring. | OpenRouterObservability | FDE Academy · M4 | queued |
| ENG-013 | LLM evaluation pipeline LLM-as-judge plus RAGAS metrics over custom domain benchmarks. | RAGASDeepEvalLangSmith | FDE Academy · M4 | queued |
| ENG-014★ | Production corrective RAG Self-correcting retrieval over 10K+ enterprise documents. | Corrective RAGRerankingRedis | FDE Academy · M5 | queued |
| ENG-015 | GraphRAG pipeline Entity extraction into a Neo4j knowledge graph with graph-based retrieval. | Neo4jGraphRAG | FDE Academy · M5 | queued |
| ENG-016 | Multi-modal RAG Unified retrieval across PDFs, images, and tables. | Unstructured.ioMilvus | FDE Academy · M5 | queued |
| ENG-017★ | Self-correcting agent LangGraph agent with planning, tool use, reflection, error recovery, and human approval gates. | LangGraphHuman-in-the-loop | FDE Academy · M6 | queued |
| ENG-018 | Multi-agent research crew 4+ coordinated agents with delegation, shared memory, and report synthesis. | CrewAIMem0 | FDE Academy · M6 | queued |
| ENG-019★ | Custom MCP server + agent MCP server for an internal API, connected to Claude Desktop and deployed. | MCP SDKFastMCP | FDE Academy · M6 | queued |
| ENG-020 | No-code agent gallery Three production workflows built in n8n, Flowise, and Dify. | n8nFlowiseDify | FDE Academy · M6 | queued |
| ENG-021★ | Full-stack AI app React + FastAPI + LLM + vector DB, containerized with Docker Compose. | Next.jsFastAPIDocker | FDE Academy · M7 | queued |
| ENG-022 | CI/CD pipeline for AI GitHub Actions with quality gates and staged dev/staging/prod deployment. | GitHub ActionsQuality gates | FDE Academy · M7 | queued |
| ENG-023 | Cloud AI service AWS Lambda + API Gateway + SageMaker with auto-scaling and cost monitoring. | AWSSageMakerTerraform | FDE Academy · M7 | queued |
| ENG-024★ | Capstone client sprint 3-week simulation: ambiguous RFP → scope → build → demo to an industry panel, with mid-sprint requirement changes. | ScopingDelivery under pressure | FDE Academy · M8 | queued |
| ENG-025 | Production hardening Monitoring, alerting, auto-scaling, and runbooks added to a previous deployment. | PrometheusGrafanaRunbooks | FDE Academy · M8 | queued |
| ENG-026 | Portfolio assembly 24 artifacts curated, GitHub optimized, five case-study presentations. | CurationStorytelling | FDE Academy · M8 | in-progress |
★ flagship — receives full case-study treatment and a node upgrade on the home network.