About

AI/ML researcher pursuing an M.Sc. in EEE at NTU Singapore. I work at the intersection of computer vision, LLM systems, and embodied intelligence — currently at Huawei Singapore on RAW image reconstruction and VLA models.

Experience

Algorithm Research Engineer Intern

Huawei Singapore Research Center

Jul 2025 — Dec 2025

RGB-to-RAW reconstruction: ReRAW model, PSNR 52.81 → 56.98, SSIM 0.91 → 0.95
LMCache on Ascend 910B: CacheBlend integration, KV hit rate surpassing vLLM prefix cache
VLA/RAW-to-Action: SpatialVLA on Bridge dataset, success rate 34.5% → 49.5%
RawAdaptationModule: learnable channel projection, bridging 4-ch RAW to RGB encoder
AgentDriver: full-stack Ascend migration, Qwen2.5-32B + MindSpeed-LLM fine-tuning
Outstanding R&D Contribution Award (team)

Chip Verification Engineer Intern

Infineon Technologies

Dec 2022 — Jul 2023

Tcl/Python automation tools, qualification runtime reduced ~33%
RTL bring-up qualification experiments with chip design engineers

Implementation Engineer Intern

Kingdee International Software

May 2024 — Jul 2024

ERP system implementation for Chongqing Xinhua Bookstore Group
Python data processing automation, efficiency improved over 70%

Education

M.Sc. Electrical & Electronic Engineering

Nanyang Technological University

Aug 2024 — Present

CGPA 4.50/5.00 — Highest Distinction
EEE Global Merit Award (21/2114)
Genetic Algorithms & ML, NLP

B.Eng. Electrical & Electronic Engineering

Nanyang Technological University

Aug 2020 — May 2024

SM2 Scholarship

Competitions

WSDM Cup — Multilingual Chatbot

Fine-tuned Gemma2-9b-it with LoRA + int4 quantization

Rank 154 / 950

LLM 20 Questions

GRPO training from scratch, multi-agent cooperative architecture

Accuracy 83%

Toolbox

Languages

PythonTypeScriptSQLTcl

ML / AI

PyTorchvLLMPEFTAscend 910BMindSpeed-LLM

Research

Computer VisionInverse ISPRAW ProcessingEmbodied AI

Engineering

GitLMCacheDockerLinux

Contact

Email diaoshengjia@gmail.com GitHub @GoDiao

Location Singapore