CareerPlan

Inference Performance Engineer - MaaS

Singapore, SGFull-time2026-11-12 → 2026-11-16

Core

Characterize real production traffic shapes for Bitdeer's MaaS platform to ensure benchmarks, capacity plans, and cost projections accurately reflect actual customer workloads.

Role type

Senior Inference Performance Engineer (MaaS)

Builds

Trustworthy capacity forecasts, unit-economics models, and canonical metric definitions for GPU-backed LLM inference services.

Domain

AI Cloud / Large Language Model Inference / FinOps

Deliverable

production ML models

Required skills

Statistical traffic profiling, LLM inference internals (continuous batching, KV-cache), load-testing tooling, GPU cost modeling, observability stack, metric definition governance

Preferred skills

Experience with vLLM, TensorRT-LLM, SGLang, Triton Inference Server, Prometheus, Grafana, distributed tracing

Technologies

Python, pandas, SQL, Spark, k6, Locust, MLPerf, Prometheus, Grafana

Responsibilities

Profile real production request streams to build statistical models of prompt/output length and concurrency; Validate internal load-testing harnesses against live production telemetry; Translate workload data into capacity forecasts and cost-per-token models; Define and maintain canonical definitions for latency percentiles and throughput metrics; Build dashboards and reports for engineering, product, and leadership on inference performance and cost trends

Seniority

Senior, hands-on IC