1. Job Overview

• Role: Fullstack AI Engineer
• Location: Ho Chi Minh City, Vietnam
• Employment Type: Full-Time and Onsite

2. About TNI Tech

TNI Tech is a deep-tech company building SightMatrix — an enterprise-grade, AI-powered Video Management System (VMS) deployed on-premise and in the cloud for smart cities, enterprise security, and intelligent transportation.

Our platform processes live video feeds from thousands of IP cameras, runs real-time AI inference (face recognition, intrusion detection, ANPR, crowd analytics, etc), and delivers instant alerts to operators — all at the edge and in the cloud simultaneously.

We are a startup and high-ownership engineering team. Every engineer owns a meaningful slice of the product and ships features that run in production environments across the country.

3. The Role

We are looking for a Fullstack AI Engineer who can build, integrate, and maintain the software infrastructure that powers our AI video analytics platform.

This is a hands-on engineering role — not a managerial or research position. You will design APIs, build web-based operator interfaces, integrate AI inference services, and manage the data pipelines that connect cameras, ML models, and end users.

You will work across the full stack: from the Python backend that ingests AI events, to the NextJS frontend where operators monitor live feeds and review alerts, to the Docker infrastructure that packages everything for on-premise deployment.

4. What You Will Do

Backend Engineering

• Design and implement REST APIs using FastAPI
• Build async, production-quality services with SQLAlchemy (async), PostgreSQL, and Alembic migrations
• Integrate Redis for real-time event streaming, pub/sub messaging, and task queuing between services
• Manage file storage (snapshots, recorded video segments) via MinIO
• Design and maintain database schemas including vector columns and partitioned tables for multi-tenant, high-volume event data
• Implement JWT-based authentication, RBAC, and API security patterns

AI Pipeline Integration

• Integrate with AI inference services (NVIDIA Deepstream and Triton Inference Server)
• Build vector search features using Qdrant
• Develop Python services in the AI camera pipeline
• Work with video streaming protocols (RTSP, WebRTC)

Frontend Engineering

• Build responsive, real-time operator dashboards in Next.js (App Router, TypeScript)
• Implement live event monitoring with WebSocket push notifications

DevOps & Deployment

• Write and maintain Docker Compose configurations for multi-service deployments across hardware targets
• Support deployments in Center/Site distributed topology
• Write and maintain deployment, backup, and release automation shell scripts
• Manage NGINX reverse proxy configuration

5. Requirements

Must Have

• Python: 2+ years production-grade experience with Python
• AI: Familiarity with Deep learning model training and inference
• NVIDIA stack: Familiarity with CUDA, TensorRT, DeepStream, and Triton Inference Server
• FastAPI: Hands-on experience building production REST APIs with FastAPI or equivalent async framework
• SQLAlchemy: Experience with SQLAlchemy 2.x ORM, async sessions, relationship mapping, and Alembic migrations
• PostgreSQL: Proficient in PostgreSQL — schema design, indexing, query optimization, constraints
• Vector Search: Familiarity with vector databases (Qdrant, Weaviate, Pinecone, Milvus, or pgvector)
• MinIO / S3: Experience with object storage APIs for file upload, retrieval, and presigned URLs
• Docker: Can write Dockerfiles and multi-service docker-compose.yml from scratch
• Git: Strong git workflow practices (branching, meaningful commits, code review)
• Testing: Writes unit and integration tests with pytest, pytest-asyncio, pytest-mock
• Linux: Comfortable with Linux shell, systemd, cron, and bash scripting
• Communication: Can clearly explain technical decisions in writing and in code reviews
• English: Fluent in English (oral and written), at least 700 TOEIC or equivalent

Strong Preference

• Redis: Experience with Redis pub/sub, Streams, or task queuing (Celery/dramatiq/ARQ)
• gRPC: Has worked with gRPC + Protocol Buffers in Python
• Async patterns: Comfortable with Python asyncio, async/await, async generators
• pgvector / HNSW: Has used pgvector for ANN search in PostgreSQL
• Next.js: Experience with Next.js, Server/Client Components, route groups
• WebSocket: Has built real-time features using WebSocket in both backend and frontend

Nice to Have

• RTSP / WebRTC: Experience with real-time video streaming protocols
• ONVIF: Familiar with ONVIF camera protocol
• Embedded Linux: Experience with embedded Linux hardware like Raspberry Pi, NVIDIA Jetson, Qualcomm
• Hybrid architecture: Experience with hybrid architecture where AI inference is performed on edge devices and results are synchronized to the cloud
• Multi-tenant systems: Has designed systems with data isolation at the organization/tenant level

6. What We Value

• Ownership over handoffs. We expect engineers to take a feature from requirements to production — including writing the migration, the API, the UI, the Docker config, and the deployment script
• Clarity in code. Code is read far more than it's written. We value descriptive naming, typed interfaces, and well-structured service layers over clever one-liners
• Pragmatic engineering. We ship real products to real customers. We balance correctness with delivery speed and know when “good enough now” beats “perfect later”
• Cross-layer thinking. The best solutions on our team come from engineers who can reason from the database schema up to the browser interaction — not those who stay in one layer
• Bias for debugging. Our production environments run on Linux hardware with limited connectivity. We need engineers who are comfortable with docker logs, tcpdump, and reading stack traces — not just engineers who write features

7. What We Offer

• Competitive monthly salary with project-performance-based bonuses
• Social, health, and unemployment insurance
• 14 paid leave days per year
• Work on a real AI product in production — not internal tools or CRUD apps
• A small, senior team where your architectural decisions have real impact
• An open environment to work, learn new technologies, and build your experience

8. Hiring Process

• Send your application to job@tnitech.co with the title [Fullstack AI Engineer] Your Full Name
• CV Scan — CV review (3–5 business days)
• Home Test — Practical take-home test (schedule at your convenience)
• Online Interview — 1h home test review + technical background + system design
• Offline Interview — 30min culture fit + role discussion with engineering team
• Offer