[X: 001]
Production Ready

AI PRODUCT ENGINEER / FOUNDING ENGINEER

Chad Avalon
AI Product Engineer

I build AI products that behave reliably in the real world: agentic systems, tool-using LLMs, streaming UX, observability, and unit economics. My work focuses on turning ambiguous user intent into measurable outcomes—acceptance, retries, time-to-first-useful-output, and cost per successful outcome.

Agentic Systems LLM Observability Cloud Infrastructure Full-Stack TypeScript
GitHub LinkedIn Resume
Chad Avalon headshot
Product Development
Tool Calling & ReAct Patterns
RAG & Streaming UX
Observability & Unit Economics

AI Product Engineering

Click cards for details
REF_ROUTLY_ROUTER

LLM Infrastructure

Routly // Carbon-Aware LLM Router

Founding Engineer · Jan 2025 – Present

Carbon-aware LLM routing system optimizing across carbon intensity, latency, cost, and model quality using live grid signals. Production streaming with fallback chains and zero data loss.

  • + ~40% carbon intensity reduction per session
  • + Multi-region routing with vLLM backends
  • + 8.5% fallback rate, zero data loss

Engineering Challenge

Balancing real-time carbon grid signals with latency SLAs across distributed regions

REF_ROUTLY_DASH

Observability Platform

Routly Dashboard // Metrics & Observability

Founding Engineer · Jan 2025 – Present

End-to-end observability platform for routing decisions, latency (p50/p95), throughput, error rates, and per-request carbon accounting with historical persistence and real-time dashboards.

  • + Real-time latency & throughput metrics
  • + Per-request carbon accounting
  • + Historical persistence & dashboards

Product Metrics

p50/p95 latency, requests/sec, error rates, carbon grams per request, cost per successful outcome

REF_SOMNISCOPE

Agentic Analytics

SomniScope // Agentic CPAP Analytics

Independent Project

Agentic AI system analyzing high-frequency CPAP data (25Hz; ~21.6M samples/month) with RAG over personal sleep journals. ReAct-style reasoning with deterministic tool execution prevents hallucinated metrics.

  • + Tiered analytics: SQLite + DuckDB + Parquet
  • + SSE streaming with agent state visibility
  • + Langfuse observability with provenance

Engineering Challenge

Correlating subjective journal entries with objective 25Hz physiological streams at interactive speeds

REF_WHATSCAL
📱💬

WhatsCal

Conversational Scheduling

Tool-Calling Agent

WhatsCal // AI Scheduling Assistant

Independent Project

Tool-calling conversational scheduling assistant over WhatsApp/SMS for dental practices. AI-driven broadcast optimization with safety guardrails and human handoff protocols.

  • + Booking, rescheduling, FAQ handling
  • + Send-time & channel optimization
  • + Arize Phoenix observability

Product Metrics

Fill rates, time-to-fill, message success rates, handoff frequency, cost per booking

REF_CLASSIFIED
CLASSIFIED
DOCUMENT CLASSIFIED
CLICK TO DECLASSIFY

███████ █████████

██████ // ███████████ ███████

█████████ ████████

  • + █████████ ███████ █████████
  • + ████████████ █████████ ███████
  • + █████ █████████ ████████████

██████ █████████

██████████ █████████████ ████████

Professional Experience

😴

Somnology / Lead Frontend Engineer

Bay Area · Jun 2023 – Feb 2025

Led development of physician-facing sleep apnea management platform (React, TypeScript, Node.js), integrating data from 7 wearable device types. Built CPAP compliance visualizations reducing patient onboarding by 35%.

Challenge

Unifying 7 disparate wearable data formats into coherent clinical insights

Zulily

Zulily / Software Engineer

Seattle · Dec 2019 – Nov 2022

Built vendor onboarding portal (React/Redux) reducing product listing time by 150%. Improved AWS logging and error visibility across 4 teams, reducing MTTR by 70%.

Metrics

150% faster listings, 70% MTTR reduction

Syndio

Syndio / Software Engineer

Santa Cruz · Oct 2017 – Jun 2019

Implemented pay equity calculator contributing ~$100K in new sales within three months. Built legal analytics dashboards cutting review time by 40%.

Metrics

$100K new sales, 40% faster legal reviews

Montaia

Montaia Global / Frontend Engineer

Contract · 2016 – 2018

Built event and lodging management platform in React/Node.js. Increased event page engagement by 300%.

Metrics

300% engagement increase

Systems & Toolchain

STACK

Core Stack

  • TypeScript • Node • Next.js
  • Python • Django • FastAPI
  • Rust workers • Wasm
  • LangChain • Custom ReAct loops
  • DuckDB • Parquet • Redis
OBSERVE

Observability & Infra

  • AWS • GCP • Fly.io orchestration
  • OpenTelemetry • Grafana • Langfuse
  • Streaming SSE • WebRTC telemetry
  • Carbon-aware schedulers
  • MTTR sims & chaos drills
ECON

Product & Economics

  • Agent cost tracing & FinOps reviews
  • Unit economics instrumentation
  • Experiment design & eval harnesses
  • Compliance-ready knowledge graphs
  • Stakeholder-ready launch narratives

About & Approach

Site-walk style discovery, blueprinting agent responsibilities, and aligning telemetry budgets before code lands. I coach teams through launch windows, deliver playbooks, and stay through the first regression to make sure the system defends itself.

Education

M.S. Computer Science — Sofia University

M.A. Integral Psychology — JFK University

Modalities

Embedded residencies

Fractional product/eng leadership

Advisory boards & diligence

COORDINATES

CURRENT GRID LOCK

LAT: 37.7749 // LONG: -122.4194

Operating between SF & Oakland. Available for onsite architecture weeks or async deep-work sprints with full playback artifacts.

// Contact

cavaloni@aiprodeng.studio

Download resume

// Signal / Telegram upon request