AI ENGINEER - Part-Time

Role Overview

We are seeking a Technical Lead to own the end-to-end strategy, architecture, and operations of a production-grade LLM inference platform. This role spans commercial partnership management, distributed systems engineering, and platform automation — owning everything from GPU cluster architecture to go-to-market execution on third-party marketplaces.

Key Responsibilities

Platform & Partnership Leadership

• Lead end-to-end launch strategy for LLM inference channels on external marketplaces (e.g., OpenRouter), including business case development, pricing strategy, and partner relationship management
• Own vendor and partner relationships, including routine reviews and pricing negotiations for the inference stack
• Scale channel throughput to production volumes exceeding 10B+ tokens/day

Infrastructure & Platform Engineering

• Design and build custom Kubernetes Operators (CRDs + controllers) to manage LLM inference deployments and benchmark runs as declarative, first-class cluster resources
• Automate model provisioning, GPU placement, and scaling so new models move from config commit to serving traffic without manual intervention
• Architect and operate multi-node distributed inference deployments over InfiniBand
• Implement KV-cache reuse and disaggregated prefilling (e.g., LMCache) to drive significant throughput gains (1.6×+)
• Build GPU autoscaling systems (e.g., KEDA-based) tied to real-time load signals

Performance & Capacity Planning

• Build reproducible benchmarking systems (e.g., on NVIDIA aiperf) exposed as first-class CRDs across the GPU fleet
• Define and measure sustainable throughput per (model, chip) under strict latency SLAs (e.g., TTFT < 5s)
• Translate benchmark data and marketplace pricing into break-even tokens/day and required per-replica TPS to drive GPU capacity planning decisions

Observability & Operations

• Build observability systems covering TTFT, TPS, and per-API-key latency across the fleet
• Design and deploy AI-driven operations agents (built on in-house multi-tenant agent runtimes) for automated anomaly detection and root-cause analysis, integrating Prometheus triggers and custom MCP tools
• Drive incident resolution at scale, resolving hundreds of production incidents

Qualifications

• Proven experience architecting and operating distributed GPU inference infrastructure at production scale
• Hands-on expertise with Kubernetes Operator development (CRDs, custom controllers)
• Deep familiarity with LLM serving optimization techniques (KV caching, disaggregated prefilling, autoscaling)
• Experience with performance benchmarking and capacity planning for GPU workloads
• Track record of owning commercial partnerships and pricing strategy for technical platforms
• Experience building or integrating AI agent systems for operational automation
• Strong cross-functional leadership: comfortable operating across infrastructure, product, and business strategy.

Important Note:

> Please share your resume in word format with [email protected]

> Important Note: If this requirement is not a match for you please refer to your friends.

> Interested professionals can reach out to me for Confidential Discussion @ +65- 9060-4050.

Best Regards,

Dilip Kumar Daga

Vice President - Strategic Accounts

Helius Technologies Pte Ltd

36, Robinson Road,#13-05, City House, Singapore 068877

DID: +(65) 6429-9407

Mobile: +(65) 9060-4050

Fax: +(65) 62222213

Email id: [email protected]

http://helius-tech.com

Registration No : R1108376

EA Licence No : 11C3373

https://www.linkedin.com/in/dilipdaga/