Updated July 6, 2026

Alec Maliwanag

Senior Software Engineer · AI & agent systems

I build the operator layer for AI-assisted engineering — orchestration that runs fleets of coding agents, and verification that proves their work.

Nine years shipping SaaS and marketplace platforms from seed to Series D. Now building AI agent systems for private SaaS clients, and open to full-time senior engineering roles.

Proven record helping companies grow

Seed through Series D — now independent

Seed2021–2022Tab For A Cause
Series A2023–2024Complete.so
Series B–D2020–2023Upside
Series D2017–2020HotSchedules
Now2026–NowIndependent

i.

Projects

What I'm building now.

Live products, open-source operator systems, and applied-AI work — each with a memo, demo, or repo. The full employment record lives in the resume, one click above.

Graph-native index

Explore projects, services, workflows, proof artifacts, and source-backed outcomes as a connected system. Symphony control-plane and Harmony Mobile detail lives in the Harmony memo.

Open graph canvas

01

Verifiable agent-build evidence

Proof-grade demo; attestation core + independent verifier shipped

NativeLink Agent Flight Recorder

When can you actually trust an AI agent's work? NLFR records an agent's build as signed, tamper-evident evidence and ships a standalone verifier that checks it. Trust comes from the evidence itself, not the agent's word — and the moment a single byte is altered, verification fails.

My strongest applied-AI proof: a tool that records what an AI agent actually did as signed, tamper-evident evidence, so its work can be independently verified instead of taken on trust.

Read project memo Source

02

Operator system for agent fleets

Open-source operator system; public-safe demo media shipped

Harmony

The operate half of running fleets of coding agents: Fleet, a typed Go control plane (its CLI is named symphony), runs and records agent work, while Operator Cockpit, an Electron/React surface, makes that work visible, reviewable, and controllable — with preview-first actions that keep mutation authority behind Fleet-owned gates.

The operator layer for AI-assisted engineering: one desktop app to run fleets of coding agents, preview risky actions before they happen, and keep an audit trail of everything — grounded in 900+ private fleet runs.

Includes Symphony + Harmony Mobile as supporting surfaces.

Read project memo Source

03

Coffee product suite

Live site recorded; proof diagrams approved

Turtle Rock Coffee

A full-stack coffee product suite connecting Taste, Brew, Library, and Azulejo / You across web, API/data, and native mobile companion work.

Shows full-stack product range: a connected coffee suite across web, native mobile, and API/data, built as one system rather than four separate apps.

Read project memo Live site

04

React design-improvement package

Public demo, package, and docs

Open Loop UI

Open Loop UI is an in-app design improvement loop for React: point at real UI, describe the change, preview the route, and hand the request to a local CLI, agent, issue tracker, or PR flow through a small JSON contract.

A published React tool that lets someone point at any part of a live app, describe a change, and hand it straight to an AI or a developer to build.

Read project memo Live site Source npm

05

Fact-safe job-search pipeline

Active, operator-run daily

Job Pipeline

A second domain on the same Symphony orchestration substrate: ingest, triage, research, fact-safe auto-draft, independent audit pass, apply-packet release, and a truthful outcome learning loop the operator runs daily. V3 ships a structural defense against hallucinating AI-apply tools: tailored résumé content is validated against the base-resume corpus before any PDF renders, making fabrication architecturally impossible.

An AI job-search pipeline built so it can't lie: every generated résumé is checked against a verified fact base before it ships, and a human sends every application. 1,228 leads triaged, 16 audited apply packets.

Read project memo

ii.

Work modes

Full-time first, scoped work real.

Full-time senior SWE remains first in positioning. Upwork, freelance, and scoped advisory are first-class support and traffic-inflow channels when the work is bounded, evidence-backed, and useful to the same product/platform story.

01 / Primary

Senior IC / founding engineer

Full-time product and platform engineering for teams that need senior execution across product surfaces, backend systems, data workflows, and applied AI features.

01TypeScript, React, Next.js, Python/FastAPI, Node, Go
02Backend product/platform systems and cloud architecture
03Applied AI, LLM product work, agents, evaluation, orchestration

02 / First-class inflow

Upwork / scoped product build

A real inbound path for bounded freelance or Upwork-style builds when the problem, constraints, timeline, and handoff are crisp enough to make senior execution useful.

01React/Next.js interfaces and product workflows
02Python/FastAPI, Go, Node, GraphQL, or integration services
03Clear milestone, delivery shape, and production handoff

03 / Support

Architecture / AI advisory

Focused support before a build, migration, agent workflow, or data integration burns budget in the wrong direction.

01System design and migration sequencing
02LLM feature scoping and evaluation plans
03Build-versus-buy and operational risk review

Full-time senior engineering is the priority. When a problem is genuinely bounded, the scoped-build route lays out fit, proof, risks, and the smallest useful milestone before it becomes real work.

Review scoped-build fit

iii. / About

Product engineering with systems depth.

I am an AI engineering consultant and senior / founding software engineer based in the Bay Area, with 9 years building and scaling SaaS and two-sided marketplace platforms. My strongest fit is senior individual-contributor or applied-AI work with clear scope, deep TypeScript and Python needs, and enough product context to make engineering judgment matter.

My work spans React and TypeScript product surfaces, Python and Go backends, GraphQL APIs, serverless data pipelines, cloud cost and reliability work, and applied AI / LLM product systems. I have worked in startup environments where the useful answer is rarely just code; it is code plus sequencing, tradeoffs, and operational taste.

Recent source-backed outcomes include helping Complete.so scale ARR 4x to $1mm+, reducing annual AWS costs by $300k and enabling real-time billing on $700M+ in-app transactions at Upside, reaching 2M+ users with AI-generated offers, and leading a Tab For A Cause rewrite that improved growth, acquisition cost, revenue per user, and attrition.

Since March 2026 I have run an independent AI-engineering practice for private SaaS clients — production agentic workflows and verification harnesses where every AI-authored feature ships with end-to-end coverage and a recorded walkthrough under human review. The open-source half of that practice is public: Harmony, an agent-fleet operator system with a desktop cockpit, and NLFR, a flight recorder that makes an AI agent's work independently verifiable.

iv.

Toolkit

Skills kept close to source.

The stack list favors canonical resume skills and active project-system work over an inflated technology cloud.

AI / agents

Claude Code
Claude
MCP
Agent orchestration
Agent harnesses
Evals
LLM workflows
Context engineering
Cursor
OpenAI

Backend / data

Python
FastAPI
Go
Node
Java
GraphQL
Postgres
MongoDB
Snowflake
Redis

Frontend product

TypeScript
React
Next.js
Playwright
Jest
Relay
Redux
CSS

Cloud / infra

AWS
S3
Lambda
SNS/SQS
ECS
Kubernetes
Terraform
Docker
CI/CD
Vercel

v. / Now

Current focus.

Updated July 6, 2026

01Consulting for private SaaS clients on production agent systems, AI verification harnesses, and agent-fleet orchestration, all under human review.
02Open to full-time senior software engineering roles focused on TypeScript, Python/FastAPI, React, Node, Go, backend product/platform engineering, and applied AI / agent systems.
03Building the open-source half of the practice in public: Harmony (agent-fleet operator system) and NLFR (verifiable agent-build attestations).
04Based in Bay Area, CA.

vi. / Contact

Let's build something real.

I am focused on full-time senior software engineering roles, with Upwork, freelance, and scoped builds treated as real inbound channels when the work is concrete, source-backed, and useful.

amaliwanag@gmail.com Review the work