TypeScript linting hits Go speed as Terraform AWS provider, Netflix Conductor and NVIDIA PAIR reshape infra tooling

tsgolint v7 lands type-aware TypeScript linting through typescript-go

On September 11, 2026, the tsgolint project reached a stable v7, bringing Go-powered type-aware linting to Oxlint. Under the hood, tsgolint reuses TypeScript's semantic analysis through the typescript-go compiler, while Oxlint continues to handle configuration and file discovery. The release targets compatibility with TypeScript 7.0.2 and already covers 59 of 61 type-aware rules, with reported performance gains over ESLint. For teams running large TypeScript monorepos, the practical change is that type-aware lint passes no longer require a Node-hosted ESLint pipeline, which simplifies CI images and shortens feedback loops on every save.

Terraform AWS provider 6.62.0 extends coverage as AWS surface grows

On September 11, 2026, the Terraform AWS provider shipped v6.62.0. The release adds support for additional AWS capabilities and refines how Terraform inspects and reconciles pre-existing infrastructure. For practitioners, the signal is the cadence itself: a single minor version drop in one day indicates the AWS surface is changing faster than a six-week release window can absorb, so drift detection and refresh tooling inside Terraform pipelines deserves attention alongside any new resource adoption.

Netflix Conductor 4.0 reworks orchestration for 10X larger workflows

Netflix has reworked its Conductor workflow orchestration engine to support larger payloads, raising the supported workflow size from about 2,500 to 30,000 tasks and trimming p99 workflow evaluation latency by roughly 40 percent. Conductor 4.0 separates workflow metadata from task data, moves evaluation to asynchronous processing, and adds dynamic worker allocation and concurrency controls. The engine now drives about 420 million monthly workflow executions. For teams that hit Conductor's prior ceiling, this changes both planning assumptions and the worker-side tuning required to keep tail latency in check.

NVIDIA Personal AI Router beta spreads inference across local machines

On September 11, 2026, NVIDIA opened a beta of Personal AI Router (PAIR), a layer that pools inference capacity across multiple computers on a local network and routes AI requests among them. The design targets local multi-agent workloads, where many independent model calls can otherwise saturate one GPU. For practitioners experimenting with on-device agents, the question is no longer whether a single box has enough VRAM but how to keep peer machines warm and how to route to the cheapest capable host.

LinkedIn compresses a multi-teacher ranking pipeline into a 0.6B model

LinkedIn published the training recipe behind its AI-powered job search on September 11, 2026. A multi-teacher distillation pipeline compresses knowledge from large teacher models into a compact 0.6B-parameter ranking model, which LinkedIn reports trains roughly 8x faster than the prior approach. The takeaway for developers is concrete: very large teachers can produce small, deployable rankers, and the distillation harness is now a documented pattern rather than folklore.

React 19.3 ships while Copilot app, Stack Overflow Story and Form3 multi-cloud surface for practitioners

The wider developer ecosystem also moved on September 10 and 11, 2026. js, cn and Maps updates, plus an Expo Modules roundup and Navigation Benchmarks. GitHub's Copilot blog walked beginners through viewing diffs, running terminal commands and previewing web apps side by side inside the Copilot app, useful context for anyone wiring agent-generated code into review pipelines.

Stack Overflow reintroduced Developer Story, putting individual identity back at the center of Stack profiles. A Form3 architecture talk, presented on September 11, 2026, walked through running three clouds at once and the engineering tradeoffs in cross-cloud networking, CockroachDB and NATS, custom Kubernetes operators, and regional disaster recovery expectations across UK, European and US financial markets. For practitioners building across front-end, infrastructure and platform tooling, the throughline is convergence: agents and routing on one side, multi-cloud and orchestration on the other.

0's async evaluation model against their largest existing workflow. If you build or test small utilities for the browser, an Auto Counter or MIME Type Lookup pair can serve as a quick smoke target for the new Copilot preview flow.

What to verify next

Watch for the tsgolint rule-coverage counter to tick past 59 of 61 in subsequent releases, the Terraform AWS provider changelog for further drift-detection refinements, and the NVIDIA Personal AI Router beta channel for routing policy documentation. The React 19.3 release notes, the Stack Overflow Developer Story profile editor and the Form3 multi-cloud presentation are all available on demand for follow-up reading. Concrete next step: re-run your largest lint job against tsgolint v7 and your heaviest orchestration workload against Conductor 4.0, and capture the wall-clock difference before planning the next migration.