Research Engineer

Neuralk; Singapore; Hybrid; SGD 180,000 - 260,000; Full time

Benefits offered by this job

Equity (BSPCE)

State-of-the-art ML infra

Paid leave and time-off policy

Occasional remote work; offices in IP

Job summary

Neuralk is an early-stage AI startup building scalable predictive infrastructure for enterprise data science. We seek a Research Engineer to own pretraining, optimize inference latency, and write low-level GPU code to enable large-scale experimentation.

You will work at the intersection of systems and research, shaping architecture direction. The role emphasizes strong ML systems experience, Python/PyTorch expertise, and collaboration with a fast-moving team in a research-driven environment.

Qualifications

• Master’s or PhD in Computer Science, ML or related field.
• 5+ years of experience in ML systems focusing on training and inference optimization.
• Proficient in Python and PyTorch with solid software engineering practices.
• Hands-on experience writing CUDA kernels or using low-level GPU optimization libraries.
• Strong understanding of transformer architectures and scaling behavior.
• Experience with distributed training frameworks and cluster environments.
• Excellent English communication skills and ability to publish in ML venues.

Responsibilities

• Design and optimize transformer inference to reduce latency and scale production models.
• Build and maintain scalable pretraining infrastructure with distributed training and mixed precision.
• Develop CUDA kernels and low-level GPU code for bottleneck optimization.
• Collaborate with research to test architectures and reproduce experiments at scale.
• Develop internal tooling for experiments, evaluation, and benchmarking.
• Contribute to publications and engage with the ML community.

Skills

Python

PyTorch

CUDA kernels

Distributed training

Transformer models

English communication

Research publication

Inference serving

Education

Master’s or PhD in CS/ML

Tools

DeepSpeed

FSDP

SLURM

PyTorch

TensorRT

ONNX

Triton

cuBLAS

CUTLASS

Job description

Neuralk is a deep-tech company building the next generation of Foundation Models for Data Science. Our mission is to build the predictive layer for businesses, transforming data science from a series of one-off initiatives — stitched together across silos, overly bespoke, and dependent on a handful of specialists — into a durable capability: a scalable predictive infrastructure that continuously learns from an organization’s data and powers decisions across the enterprise.

As an early-stage, well-funded AI startup, Neuralk builds on state-of-the-art research to solve concrete business challenges. We value clarity over complexity, strong fundamentals over hype, and fast iteration grounded in rigorous engineering. Joining Neuralk means working hard in a fast-moving, research-driven environment, with a high level of ownership and the opportunity to shape a core product at the intersection of machine learning, engineering, and real-world impact.

About the Role

Scaling foundation models to structured data is not a solved problem. Unlike text or vision, tabular data has no canonical tokenization, no natural sequence, and no shared feature space across datasets, which means standard scaling laws, positional encodings, and pretraining objectives don’t transfer. The architectural and optimization challenges are fundamentally different, and largely open.

At Neuralk, we’re building the infrastructure and the models to solve this. As a Research Engineer, you’ll work at the intersection of systems and research — owning the pretraining stack, pushing inference to production-grade latency, and writing the low-level GPU code that makes large-scale experimentation possible. The same people who build the systems shape the research direction.

Role & Responsibilities

• Transformer inference optimization: Design and implement optimizations to reduce latency and improve scalability of our foundation models in production: attention mechanisms, KV-cache, batching strategies, quantization.
• Pretraining optimization: Build and maintain the training infrastructure that enables us to scale pretraining efficiently (distributed training, mixed precision, gradient checkpointing, throughput optimization)
• CUDA kernel development: Write low-level GPU kernels to optimize critical bottlenecks, from custom attention variants to data preprocessing pipelines.
• Architecture experimentation: Collaborate with the research team to implement, train, and evaluate novel architectures adapted to structured data at scale.
• Tooling & infrastructure: Develop the internal tooling (experiment tracking, evaluation pipelines, benchmarking) that enables rapid and reproducible research iterations.
• Research contribution: Contribute to publications and engage with the broader ML research community.

Profile

• Master’s or PhD in Computer Science, Machine Learning, or a related field
• 5+ years of experience in ML systems, with a strong focus on training and inference optimization
• Deep proficiency in Python and PyTorch, with strong software engineering practices
• Hands‑on experience writing CUDA kernels or using low‑level GPU optimization libraries (Triton, cuBLAS, CUTLASS)
• Solid understanding of transformer architectures, training dynamics, and scaling behavior
• Experience with distributed training frameworks (DeepSpeed, FSDP, or equivalent) and cluster environments (SLURM)
• Excellent communication skills in English
• Self‑starter, autonomous, and comfortable in a fast‑paced startup environment
• Publication record at top ML venues (NeurIPS, ICML, ICLR, MLSys)
• Experience with inference serving frameworks (vLLM, TensorRT, ONNX)
• Contributions to open‑source ML infrastructure
• Experience with structured or tabular data

What We Offer

• Equity (BSPCE), to reflect the value you bring to Neuralk and to foster a shared journey
• Substantial compute resources and state-of-the-art ML infrastructure
• French level paid leave and time‑off work
• Dynamic work setting. Although our preference is for in‑person collaboration, we will be flexible with occasional remote work arrangements. Offices in Paris and London.