Reljod T. Oreta

Capabilities & Specialization

What I Do

I build systems that think, reason, and execute continuously in production.

Remote VPS Agent Architecture 24/7 ACTIVE

PERSISTENT MEMORY Honcho Vector Sync

UNIFIED GATEWAY Discord / Telegram Bot

AUTONOMOUS OPS PR Review Cron & Sweeps

SECURITY HARDENING 1Password Secret Injection

Zero local dependency • Fully autonomous execution while you sleep

Personal Autonomous Agents (100% Remote via VPS)

Dedicated Linux VPS Hermes Agent Honcho Memory Discord Bridge Google Workspace API Isolated Worktrees

⚡ 100% Remote Turnkey Deployment: I architect and launch dedicated personal AI agents hosted entirely on private Linux VPS instances. You don't need your laptop open or local daemons running.
⚡ Persistent Memory & Context: Integrated with Honcho and vector search to retain episodic and working memory across weeks of operations, tools, and conversations.
⚡ Autonomous Code Operations: Scheduled PR review sweeps (`pr_sweep.py`), automated testing in isolated git worktrees, and strict secret-isolation guards that never leak credentials.
⚡ Daily Life Integrations: Real-time Google Calendar push webhooks (calwatch), Google Drive ingest pipelines, and multi-channel messaging bots.

Quantitative Eval Engine DEEPEVAL DAG

[Pipeline] Input Screening: 97% Ineligible Filtered

[DAG Eval] Factual Consistency: 0.984 / 1.000

[DSPy] Prompt Optimization: Converged

[LangFuse] Span Tracing: 100% Monitored

Consensus-driven topologies • Zero prompt hallucination

Enterprise Multi-Agent LLM Orchestration

LangGraph CrewAI DSPy DeepEval LangFuse AWS Bedrock Pydantic AI

⚡ Multi-Agent Consensus Topologies: Designing coordinated agent teams with specialized personas (research, code review, data verification, synthesis) rather than fragile monolithic prompts.
⚡ Evaluation-Driven Reliability: Rigorous evaluation harnesses using DeepEval custom DAG metrics and DSPy data-driven prompt optimization to ensure outputs stay factual and bounded.
⚡ Upstream Eligibility Filters: Designed Lambda pipelines screening out 95 to 97% of ineligible inputs before LLM inference, eliminating downstream compute and token costs.

FinTech Scale Metric -90% OPEX

Maya Philippines AI KYC Engine:

• Operating Cost Reduction: 90%

• Automated Approval Lift: +7 points

• Verification Automation: +8 points

Distributed across AWS Lambda, EKS, EventBridge & Databricks

High-Scale Distributed Backends & Cloud Infra

Python FastAPI Next.js Node.js / Fastify AWS Lambda & EKS Celery PostgreSQL & Aurora

⚡ Proven FinTech Scale: Core engineer on Maya Philippines user profiles, replacing expensive vendor KYC with an in-house AI engine serving tens of millions of users.
⚡ Fault-Tolerant Async Pipelines: Resilient background worker systems on Docker, EKS, Modal, and Trigger.dev that retry and recover long-running jobs without dropped transactions.
⚡ Robust Cloud Engineering: End-to-end Terraform infrastructure, GitHub Actions CI/CD, database migrations, and zero-trust microservice architecture.

Executive Agent Brain Jod OS / Hermes

24/7 autonomous agent connected to Discord channels, Telegram, and mobile web clients. Manages subagents, executes commands, schedules jobs, and delegates multi-agent tasks continuously.

Stack: Hermes • Python • Discord API Active 24/7

Autonomous PR Pipeline pr_sweep.py + Webhook

Automated webhook and 30-minute cron sweeps that inspect every pull request across private repos. Runs in isolated git worktrees with strict secret scanning and policy enforcement.

Stack: Git Worktrees • Antigravity • GitHub Automated

Persistent Memory Core honcho.freed.kita-agents.dev

Self-hosted Honcho memory instance synchronizing working and episodic memory across sessions, conversations, and agent tools so the system remembers context indefinitely.

Stack: Honcho • PostgreSQL • Vector Embeddings Synced

Workspace Automation calwatch + Drive Ingest

Receives push webhooks from Google Calendar for instant schedule tracking and executes automated Google Drive media synchronization for automated clipping and asset storage.

Stack: Google Workspace API • Webhooks • Python Real-Time

Zero-Trust SSO Gateway login.freed.kita-agents.dev

SuperTokens-backed reverse proxy gate in front of internal tools, allowing seamless single sign-on across subdomains while hard-blocking unauthorized signups and isolating secrets.

Stack: SuperTokens • FastAPI Gate • Caddy Hardened

Private Tooling Ecosystem *.freed.kita-agents.dev

Production suite including Excalidraw whiteboards, Quartz knowledge garden (notes), Applyd automated job tracker, Securo financial ledger, and HyperFrames video editor.

Stack: Docker Containers • Systemd • Caddy Self-Hosted

Want Your Own 24/7 Agent Running on a Private VPS?

I replicate, customize, and deploy this exact turnkey personal autonomous agent stack for founders, executives, and engineers. 100% remote on your private VPS, with zero local laptop dependencies.

Dedicated 24/7 Runtime: Runs continuously on your private cloud VPS with persistent memory (Honcho).
Multi-Channel Gateways: Message your agent through private Discord channels, Telegram, or mobile web app.
Automated Workflows: Calendar schedule tracking, email routing, autonomous PR review, and daily briefings.
Zero-Leak Security: Credentials injected through 1Password or encrypted vaults; your keys never leak to logs.

Connect on LinkedIn

// HOW WE WORK TOGETHER

1. Architecture Discovery: We map your daily workflows (calendar, email, GitHub, notes).

2. VPS Provisioning: I configure an isolated Ubuntu VPS, Docker runtime, and Caddy TLS.

3. Agent Tuning: Connect your models, memory database, and Discord/chat bridge.

4. Handover & Hardening: Fully hardened, documented, and operating 24/7.

Career History

Work Experience

7+ years delivering high-impact engineering across FinTech, AI SaaS, and enterprise consulting.

AI Consultant

Acquire Intelligence

Jul 2026 — Present

Remote / Hybrid

Advising enterprise leadership on multi-agent LLM systems, autonomous tool-use protocols, and model evaluation architectures. Designing custom DAG-based evaluation harnesses with DeepEval and data-driven prompt optimization with DSPy to ensure production agent pipelines operate without hallucinations.

• Impact: Production multi-agent consensus topologies & quantitative evaluation standards

LangGraph CrewAI DSPy DeepEval AWS Bedrock FastAPI

Full Stack Python Engineer / AI Engineer

Cloud Employee

Dec 2025 — Jun 2026

Remote

Engineered agent features for a multi-tenant AI SaaS platform on Python FastAPI, TypeScript, Next.js, and PostgreSQL. Designed a multi-agent eligibility pipeline on AWS Lambda that screened out 95 to 97% of ineligible records before report generation, eliminating unnecessary downstream LLM inference costs. Scheduled asynchronous agent workflows across Celery workers on Docker and AWS EKS.

• Impact: 95-97% upstream record filtering, saving major LLM inference spend

CrewAI LangGraph Celery Docker AWS EKS LangFuse Next.js

AI Engineer

Spark Intelligence, Inc.

May 2025 — Nov 2025

Remote

Constructed full-stack multi-tenant AI SaaS on Next.js, TypeScript, and Supabase, with agent logic in LangChain and LangGraph across OpenAI GPT, Google Gemini, and xAI Grok. Engineered fault-tolerant inference jobs on Modal and Trigger.dev that recovered long-running workloads without dropping tenant transactions.

• Impact: Horizontally scalable, zero-drop inference pipelines with DSPy & DeepEval verification

Modal Trigger.dev LangChain Supabase Gemini Grok

Senior Software Engineer

Maya Philippines, Inc. (Digital Bank & Wallet)

Mar 2022 — Apr 2025

Metro Manila, Philippines

Core engineer on the User Profiles team of the leading digital bank in the Philippines. Partnered with data science to replace a costly third-party KYC vendor with an in-house automated AI engine. Built high-scale streaming and data warehousing pipelines on AWS Firehose, S3, Aurora, Lambda, and Databricks.

• Impact: Cut KYC operating cost by 90%, +7 points approval rate, +8 points automation rate

FastAPI Fastify Java Spring Boot AWS Lambda EventBridge Databricks Aurora

Software Engineer III

FactSet Philippines, Inc.

Feb 2020 — Mar 2022

Taguig, Philippines

Led Revere finance data integration, migrating legacy Perl and C++ pipelines to modern C# ASP.NET with Entity Framework, MSSQL, Vue.js, and Python Flask. Automated manual data ingestion tools.

C# ASP.NET MSSQL Vue.js Python Flask PostgreSQL

Prototype Engineer

Yuro Product Development and Design Solutions

Jun 2019 — Feb 2020

Philippines

Built an end-to-end computer vision prototype for tennis match line officiating using OpenCV, YOLO, and PyTorch. Architected AWS IoT Core solutions over MQTT and created interactive Raspberry Pi kiosk prototypes.

OpenCV YOLO PyTorch AWS IoT Core MQTT Raspberry Pi

Licensure & Honors

Achievements & Certifications

National board exams, national scholarships, and university engineering representation.

⚖

Licensed Electronics Engineer (ECE)

Philippine Professional Regulation Commission (PRC)

National Board Exam Passer • 2019

⚙

Licensed Electronics Technician (ECT)

Philippine Professional Regulation Commission (PRC)

National Board Exam Passer • 2019

★

DOST National Engineering Scholar

Department of Science and Technology (DOST-SEI), Republic of the Philippines

5-Year Full Merit Scholarship • 2014 — 2019

🎓

BS Electronics Engineering

Polytechnic University of the Philippines • Manila

Graduated 2019

🏁

Shell Eco-Marathon Asia

Singapore • Represented PUP Electronics Engineering Team

International Energy Efficiency Competition • 2017

Interactive Terminal

Agent Console // freed-cli

Query my profile, live VPS setup, and production metrics right from the shell.

reljod@freed-node-vps: ~/profile (interactive) PHT UTC+8

visitor@me.freed:~$ whoami

RELJOD T. ORETA — AI Consultant & VPS Autonomous Agent Architect
Currently: AI Consultant @ Acquire Intelligence (Jul 2026 - Present)
Specialty: 24/7 personal autonomous agents on VPS, multi-agent LLM systems, and high-scale backends.
Type "help" to see available terminal commands or click the shortcut chips below.

visitor@me.freed:~$

Quick commands:

Hi all, I'm Reljod 👋