- Project
- Hybrid AI Platform.
- Delivery Model
- Scrum@Scale
I moved into a new AI division focused on product development. It needed service developers who could work closely with ML Engineers, Data Scientists, Hardware Teams, Consumers, and at times external partners.
I was a developer in one Scrum team, building my own services, and co-led the delivery of another squad that owned other AI services.
Impact
- Owned and built 6 OpenAI-compatible AI services that shipped to every Dell AI PC customer on AMD and Qualcomm hardware and powered capabilities showcased at CES and DTW. Within months of launch, every service was adopted by customers that couldn't use AI before due to privacy or costs.
- Delivered Dell's first NPU-based ASR service from zero to production in 4 months. It was followed by Dell's first NPU-based ONNX SLM chat services on both platforms in 5 months.
- Built 2 other services, for function calling, that enabled Agentic Edge AI across those Dell AI PCs. They were Dell's first ONNX-based function-calling services in production on both platforms and the default for a year, strengthening data privacy and eliminating recurring cloud costs for Dell's clients.
- In parallel, took on extra work by voluntarily researching where AI could be applied across the division, with no dedicated team, no funding, and no guarantee the ideas would clear the formal review and pay off.
Ownership
- Owned the services' research, development, testing, model validation (MMLU/WER), and deployment (referencing templates), using mostly pre-optimized models, plus a few I exported myself with vendor-specific tooling.
Cross-team Delivery & People
- Co-led the day-to-day delivery for a squad of 8 mid-level and senior developers new to Edge AI, sharing the role across time zones and escalating risks to my EM, who managed 10 squads.
- Cleared the squad's impediments and resolved cross-team dependencies and integration defects, working with 5 partner Scrum teams (ML, HW, Consumers) distributed across 4 time zones.
- Added technical detail to the squad's user stories and tasks based on discussions with the EM, PDM, and TPM.
- Onboarded 2 new developers and coached them in on-device AI, team processes, and debugging across heterogeneous AI PC fleets. Both became core members of the division.
- Applied AI
- Hybrid AI
- Generative AI
- Agile Application Development