AI Infrastructure Engineer

Kooya Inc; Taguig; On-site; PHP 781,200 - 1,339,200; Full time

Benefits offered by this job

Competitive Salary

WFH every Friday

HMO after 3 to 6 months

Mid-Shift Schedule

With Leave Credits

Cash convertible leaves

OT & Holiday Pay

Hybrid Work Set up: BGC, Taguig City

Job summary

Kooya Inc is seeking a highly technical AI Engineer / MLOps Specialist based in Taguig, Philippines. You will lead a core infrastructure migration from managed AI services to a fully self-hosted, open-source architecture on AWS. The role involves building automated data ingestion systems, managing a vector database, and ensuring the secure operation of a Large Language Model 24/7. Key qualifications include deep AI experience, AWS proficiency, and solid Python skills. Perks include a competitive salary, flexible working setup, and health benefits.

Qualifications

• Proven experience downloading raw model weights and serving them locally on GPUs.
• Hands-on experience with AWS, specifically configuring and securing EC2 VMs.
• Strong Python skills with web scraping experience.

Responsibilities

• Transition existing AI workflows to self-hosted models on AWS.
• Build and maintain automated pipelines to extract unstructured data.
• Deploy open-source models using optimized inference engines.

Skills

AI / MLOps

Cloud Infrastructure

Data Engineering / Python

RAG Architecture

DevOps

Tools

Docker

AWS

Python

PostgreSQL

FastAPI

Job description

About the Role

We are looking for a highly technical AI Engineer / MLOps Specialist to lead a core infrastructure migration. We are currently transitioning from managed AI services to a fully self-hosted, open-source AI architecture on AWS to optimize operating expenses and increase data privacy and control.

You will be responsible for the end-to-end pipeline: building automated data ingestion systems, managing a vector database, provisioning AWS GPU servers, and ensuring an open-source Large Language Model (LLM) runs securely 24/7.

Note: If your AI experience is limited to calling managed APIs (like OpenAI or Anthropic), this role is not for you. We need someone who knows how to allocate GPU VRAM, optimize inference speeds, and manage bare-metal Linux servers.

What You Will Build (Core Responsibilities)

• Infrastructure Migration: Transition existing generative AI workflows off managed cloud APIs and onto self-hosted open-source models (e.g., Llama 3, Qwen, Gemma) hosted on AWS EC2 GPU instances.
• Data Ingestion & Scraping: Build and maintain robust Python automated pipelines/scrapers that run daily to extract unstructured data from external web sources.
• Vector Database Management: Clean, chunk, and embed the extracted text into a Vector Database (preferably PostgreSQL + pgvector). Implement strict "upsert" logic to ensure daily updates do not create duplicate vectors.
• Local LLM Serving: Deploy the open-source model using optimized inference engines (e.g., vLLM, Ollama, llama.cpp). Apply quantization techniques (GGUF, AWQ) where necessary to maximize hardware efficiency and prevent Out of Memory (OOM) crashes.
• Backend Integration: Wrap the RAG (Retrieval-Augmented Generation) pipeline in a secure, high-concurrency REST API (FastAPI) to serve the frontend application.
• Cloud Security: Secure the AWS EC2 environment using proper VPC routing, IAM roles, and Security Groups.

What You Must Have (Requirements)

• AI / MLOps: Proven experience downloading raw model weights (Hugging Face) and serving them locally on GPUs. Deep understanding of LLM memory requirements (KV cache, VRAM allocation).
• Cloud Infrastructure: Hands‑on experience with AWS, specifically spinning up, configuring, and securing persistent Linux EC2 VMs.
• Data Engineering / Python: Strong Python skills. Experience with modern web scraping libraries (Playwright, BeautifulSoup, Scrapy) and handling messy HTML data.
• RAG Architecture: Strong understanding of embedding models, semantic chunking, and vector databases.
• DevOps: Highly proficient in Docker, specifically containerizing GPU‑accelerated applications (NVIDIA Container Toolkit).

Bonus Points If You Have

• Experience actively migrating off managed APIs to self-hosted Small Language Models (SLMs).
• Advanced RAG experience (Cross‑encoder re‑ranking, Hybrid Search).
• Experience with Caddy, Nginx, or similar tools for reverse‑proxying secure API endpoints and managing SSL certificates.

Perks

• Competitive Salary
• WFH every Friday
• HMO after 3 to 6 months
• Mid‑Shift Schedule
• With Leave Credits
• Cash convertible leaves
• OT & Holiday Pay
• Hybrid Work Set up : BGC, Taguig City

Please send your cv to [e-mail]