AI / Full Stack

AI Diagnostics & Support Workspace

AI-assisted diagnostics with source-linked answers, streamed responses, and controlled tool actions.

Illustrative AI support interface with cases, logs, source-linked answers and approval controls.
Illustrative interface concept · synthetic content

The Challenge

Support teams needed to connect fault reports with engineering knowledge and inspect the evidence behind suggested next steps.

My Contribution

I owned React/Next.js features from data contracts through interface delivery and production support. I built Python services and integrations for conversational and multimodal requests, retrieval, and agent-assisted diagnostics.

The Solution

The workspace brought case details, logs, images, and engineering references into one interface. Retrieval connected questions to relevant documents, while agent workflows broke down diagnostic tasks and called approved tools. Responses streamed back so long-running requests remained visible.

Engineering Decisions

Kafka/Celery/Redis processing; retries and duplicate-work protection; PII redaction, filters, rate limits and audits; LangSmith evaluations and token-cost monitoring; OpenTelemetry, Prometheus and Grafana; AKS/Helm rolling releases. Manufacturing quality and supply-chain exception triage.

What This Enabled

Support teams could inspect the source evidence behind suggested actions. Internal review workflows also connected manufacturing quality reports with supply-chain exceptions.

Related Projects

Back to Projects ↗
START A CONVERSATION

Let’s build something
that works.

Have an engineering opportunity or a project in mind?
Send me a message to discuss the work.

Email Daniel