I’m Jamin aka justjammin. I build AI platforms, agent tooling and the AI powered software and websites them. ORBTL is my personal engineering and music studio. A launch point for AI platform and forward-deployed engineering work.
Four shipped projects, followed by the experiments and engineering evidence behind them. Training infrastructure, agent orchestration, LLM evaluation and product delivery.
Queue-driven budget controller that measures LLM spend and applies safe routing and batch policies automatically.
Cost event contract → NestJS control plane → BullMQ job flows → Budget policies → Usage dashboard → Operational hardening
Back to lab ↑Trace-first incident workbench for reconstructing why an AI feature failed across requests, tools, and model calls.
Failure event schema → NestJS ingestion → OpenTelemetry traces → ClickHouse queries → Next.js investigation UI → Incident playbooks
Back to lab ↑GitHub automation that detects stale code examples and opens focused documentation repairs with evidence.
Repository scanner → Syntax-aware extraction → Example verification → Octokit change proposal → Actions workflow → Maintainer controls
Back to lab ↑Graph-based arbitration service that compares candidate model outputs and selects the safest useful response.
Decision contract → LangGraph state machine → FastAPI endpoint → Evidence and scoring → Fallback behavior → Evaluation release
Back to lab ↑Production retrieval pipeline combining vector search, BM25, transformer embeddings, and cross-encoder reranking.
Document ingestion → Qdrant index → BM25 retrieval → Semantic fusion → Cross-encoder reranking → Quality benchmarks
Back to lab ↑Fast request cache that reuses semantically equivalent LLM work while preserving freshness and safety boundaries.
Cache key contract → Fastify gateway → Redis storage → Semantic similarity → Freshness policy → Latency benchmarks
Back to lab ↑Text-to-SQL gateway that validates generated queries against schema, policy, and resource guards before execution.
Query intent contract → NestJS gateway → AST validation → Schema-aware generation → Resource and policy guards → Audit and release
Back to lab ↑Experiment workspace for versioning prompts, replaying datasets, and comparing model outcomes with reproducible evidence.
Experiment schema → NestJS API → Dataset runs → BullMQ execution → Next.js comparison UI → Reproducible reports
Back to lab ↑Provider-neutral LLM gateway with request tracing, Redis-backed controls, and operationally useful telemetry.
Provider contract → FastAPI gateway → Redis policy layer → OpenTelemetry spans → Usage and failure metrics → Deployment hardening
Back to lab ↑Typed feature-control plane for safely rolling out AI behavior by cohort, model, risk, and measured outcome.
Flag and cohort model → ConnectRPC service → PostgreSQL persistence → Next.js control plane → Rollout safeguards → Audit and operations
Back to lab ↑Durable workflow that turns source material into reviewed evaluation datasets with traceable provenance.
Dataset contract → NestJS intake API → Temporal workflows → Quality and provenance → Review UI → Export and versioning
Back to lab ↑Asynchronous document pipeline that extracts text, tables, and images for human-reviewed multimodal knowledge work.
Document intake → Celery processing → Docling/Tesseract extraction → Multimodal normalization → Next.js review UI → Export and operations
Back to lab ↑Inspectable multi-agent runtime for composing workflows, persisting state, and evaluating orchestration decisions.
Workflow contract → LangGraph.js runtime → NestJS orchestration API → PostgreSQL/pgvector state → Next.js run explorer → Evaluation and release
Back to lab ↑The world is constantly shifting, CHANGING
I'm not going to fall behind.
Adopt. Adapt. Adept.
Training pipelines, LoRA fine-tuning and evaluation, with reproducible runs and recovery paths you can inspect.
Agent harnesses, MCP integrations and orchestration that connect intent to implementation with explicit review gates.
Interfaces, APIs and delivery workflows built around the people using them. TypeScript, Python and the right tool for the constraint.
Experience across software and solutions engineering at Cars.com, and business sales at Apple. Translate the problem, work with the user and carry the solution through.
justjammin justjammin.me :: MACHINE-READABLE INDEX # PLAIN-TEXT MIRROR OF justjammin.me FOR AI AGENTS, CRAWLERS, AND HUMANS WHO PREFER IT RAW. ── ABOUT ───────────────────────────────────────────────────────── NAME .......... justjammin AKA ........... justjammin, nami, ORBTL FOUNDED ....... 2018 LOCATION ...... Chicago, IL AREA_SERVED ... WORLDWIDE HOBBIES ....... DJing, Music Production, Anime, Gaming, Programming, AI, Web Design, Content Creation PROJECTS ...... Training Platform, Astra, xoxo-eval, Legible, Domain LoRA Pipeline KNOWS_ABOUT ... AI platforms, LoRA fine-tuning, PyTorch, Ray, Databricks, Prompt engineering, Agent orchestration, MCP, Context engineering, RAG, Hybrid retrieval, Vector databases, LLMOps, TypeScript, Node.js, Next.js, Vue, Tailwind, Python, Go, C#, PHP, Symfony, REST APIs, Flask, CI/CD, Kubernetes, AWS, Marketing, Analytics, Community, Teaching I'm Jamin, aka justjammin. I'm based in Chicago, where I build AI platforms, agent tooling and the backend systems they run on. My background includes software and solutions engineering at Cars.com and business sales at Apple. I care about understanding what someone needs and building software that helps. These days, I work on training infrastructure, agent orchestration, retrieval and LLM evaluation. I want to be able to rerun an experiment, check the results and trust the system when someone uses it. I come at AI platform and forward-deployed engineering from the backend, with an eye on how people will use what I build. Adopt. Adapt. Adept. ── CAPABILITIES ────────────────────────────────────────────────── * WEBSITES & UX/UI I DESIGN AND BUILD WEBSITES WITH CLEAR NAVIGATION, THOUGHTFUL LAYOUTS, AND INTERFACES THAT FEEL GOOD TO USE. [WEBSITES · UX/UI DESIGN] * SOFTWARE & SYSTEMS I BUILD CUSTOM APPLICATIONS AND CONNECT THE TOOLS AND INFRASTRUCTURE BEHIND THEM. [APPLICATIONS · INTEGRATIONS · INFRASTRUCTURE] * AI WORKFLOWS & CONSULTING I HELP YOU DECIDE WHAT TO AUTOMATE, CHOOSE THE RIGHT TOOLS, AND BUILD AI WORKFLOWS AROUND HOW YOU WORK. [AI · AUTOMATION · CONSULTING] * CONTENT CREATION & AUTOMATION I CREATE CONTENT AND USE MY ENGINEERING SKILLS TO BUILD A ROBUST CONTENT-AS-CODE WORKFLOW. IT AUTOMATES REPETITIVE PARTS OF EDITING AND PUBLISHING, SO GETTING SOMETHING OUT TAKES LESS MANUAL EFFORT. [CONTENT AS CODE · EDITING · PUBLISHING] * DJING & MUSIC PRODUCTION I MAKE ORIGINAL TRACKS AND MIXES, WITH ROOM TO EXPERIMENT AND FIND SOUNDS I HAVEN'T WORKED WITH BEFORE. [DJING · MUSIC PRODUCTION] ── SELECTED_WORK ───────────────────────────────────────────────── - Training Platform: A working 77-intent classifier: 90.86% test macro-F1, calibrated human-review routing, and 11.71 ms CPU p95. Built on reproducible Kaggle training, full-state recovery, Ray/PyTorch execution, and Databricks reporting workflows. /demos/training-platform/ - Astra: A portable agent harness that takes work from product intent to verified implementation through five explicit gates, with a live review console and human approvals. https://astra-os.justjammin.workers.dev - xoxo-eval: A head-to-head evaluation workbench for coding agents. Compare Claude Code and Codex on isolated tasks, inspect the evidence, and bring in a blind judge. https://xoxo.justjammin.workers.dev - Legible: Writing tracker for authors, designed as a calmer alternative to existing trackers. https://writelegible.com ── LAB ─────────────────────────────────────────────────────────── - ORBTL DEFENDER — PLAYABLE ON THE ARCADE CABINET IN THE 3D LOBBY (CLICK IT). - ORBTL SHOT — DRAG-TO-THROW BASKETBALL ON THE LOBBY HOOP (CLICK THE BACKBOARD). ── FOR_AGENTS ──────────────────────────────────────────────────── LLMS.TXT ...... /llms.txt AGENTS.MD ..... /AGENTS.md SITEMAP.MD .... /SITEMAP.md INDEX.MD ...... /INDEX.md
