Nneuro

Available for new projects

I build AI features that ship — and the backend that keeps them running.

Freelance AI engineer. I take models from notebook to production: fine-tuned NLP, LLM assistants, and the event-driven APIs around them.

model inference, avg.
24 ms
faster after ONNX export
1.48×
microservices in the Blur platform
5

blur / moderation-pipeline

  1. New comment

    POST /comments · Spring Boot

  2. Event queued

    Kafka · comment.created

  3. Toxicity check

    PhoBERT · ONNX Runtime

    24 ms
  4. Result published

    Kafka · moderation.result

  5. Feed updated

    CQRS projection · Redis cache

  • Python
  • PyTorch
  • Transformers
  • ONNX Runtime
  • FastAPI
  • Gemini API
  • Spring Boot
  • Kafka
  • Redis
  • PostgreSQL
  • MongoDB
  • Neo4j
  • Keycloak
  • Docker
  • Next.js
  • TypeScript
Portrait of Pham Van Sy

About

Hi, I'm Sy.

I'm a software engineer in Ho Chi Minh City, finishing a Software Engineering degree at HUTECH, where I was named an Outstanding Student for 2024–2025.

My work sits where AI meets backend engineering: training and serving models, then building the services that keep them reliable. Before freelancing I built REST APIs for a clinical management platform as a backend intern at Amethyst Medical Vietnam.

Location
Ho Chi Minh City · GMT+7
Languages
English, Vietnamese
Availability
Open for projects

Services

What I can build for you

Three ways I usually help. Each one is backed by a system I've already shipped.

AI features for your product

Text classification, content moderation and LLM assistants — trained on your data or built on an API, served fast.

  • Model choice or fine-tuning on your data
  • Inference API with a latency budget
  • Evaluation against a simple baseline
  • PyTorch
  • ONNX
  • FastAPI

Backends & APIs that scale

Event-driven services that keep data consistent under load, with auth, caching and failure handling built in.

  • REST and WebSocket APIs with SSO / JWT
  • Kafka pipelines with outbox and saga
  • Redis caching, locks and circuit breakers
  • Spring Boot
  • Kafka
  • Redis

AI-powered MVPs

From idea to a deployed product: web app, backend and AI in one build, with one person accountable.

  • React / Next.js front end
  • Backend and AI wired together
  • Docker deploy with CI/CD
  • Next.js
  • Docker
  • CI/CD

Selected work

Shipped, not just prototyped

Case study · Social platform · 2025

Blur — a social network that moderates Vietnamese comments in real time

Toxic comments in Vietnamese slip past English-first filters. I fine-tuned PhoBERT on scraped YouTube and TikTok comments and wired it into a microservices platform so every comment is checked without slowing the feed.

  • Async moderation: Kafka → FastAPI model service → Kafka
  • Spring Boot services with outbox, saga and CQRS feed
  • Keycloak SSO, WebRTC calls, Gemini chat assistant
inference, avg.
24 ms
faster with ONNX
1.48×
microservices
5

Product demo · 3:24

Backend · E-commerce

Neuro Ecommerce API

REST backend for an online store: catalog, cart, orders, flash sales that don't oversell, and VNPay payments.

  • Spring Boot
  • MySQL
  • JWT
Source

Lab · Linux desktop

Coral — system-wide audio enhancer

Linux port of the open-source FxSound (AGPL-3): a PulseAudio passthrough backend, low-latency tuning and .deb packaging.

  • C++
  • JUCE
  • PulseAudio
Source

Process

How we'd work together

  1. 01

    Discovery call

    A short call to understand the problem. You get a written scope, milestones and a quote.

  2. 02

    Build in milestones

    Weekly demos and a shared repo, so you see progress, not surprises.

  3. 03

    Ship

    Deployed to your infrastructure with Docker, CI/CD and handover docs.

  4. 04

    Iterate

    Fixes and improvements after launch, as agreed in the scope.

Contact

Tell me about your project

Share a few details and I'll reply with questions or a proposal, usually within one business day.

phamvansy204@gmail.com
What do you need?
Budget
Timeline