Ashan Prapaharan.

Rivian and Volkswagen Group Technologies

DevOps Engineer · Build and Release

• Reduced AI-powered build failure diagnosis time by 68% by engineering a deterministic fast path in Python that pattern-matches CI failure signatures, bypassing a full Claude agent investigation and parallelizing GitLab trace and Datadog log fetching, whilst packaging the AI diagnosis into a ready-made prompt for coding agents.
• Improved developer productivity by building an active firmware release branch management system, replacing a stale, manually-updated Confluence page, giving 500+ TPMs and developers a single source of truth.
• Accelerated codebase onboarding by 35% through developing an interactive dependency graph visualizer for a monorepo with 870+ pipeline jobs and 1,500+ dependencies from GitLab CI into an explorable graph.
• Integrated Claude into GitLab CI, delivering automated code review on merge requests and severity based ranking on suggested fixes directly, increasing developer productivity and reducing reviewer time by 25%.

Ciena

Software Engineer Intern

• Architected a LLM-integrated log analysis pipeline for Apache CloudStack, summarizing system logs into human-readable diagnostics and reducing debugging time by 60%, enhancing observability and operational insight.
• Built Python and Bash automation tooling integrated with Ansible and Terraform to orchestrate deployments and backups, tooling with structured control flow, eliminating 85% of manual operational workflows.
• Led Kubernetes adoption across cross-functional teams to containerize testing environment for Ciena products, telemetry and data services, standardizing infrastructure, enhancing deployments and reducing overhead by 25%.

MPAC

Software Engineer Intern

• Reprovisioned AWS Resources to deliver $1000+ in cost savings across production systems.
• Developed Terraform modules to automate the creation and migration of 50+ EOL EC2 instances from CentOS 7 to Amazon Linux 2023, automating AMI creation, tagging and rollback verification for consistent environments.
• Deployed Python services using Boto3 to analyze, refactor, and validate AWS IAM policies, reducing policy count by 45% while enforcing least-privilege guarantees.

InnSys Incorporated

Part-time

• Designed a conversational backend service using Python and REST APIs to collect, validate, and structure customer contact data from website interactions, whilst increasing inbound sales outreach by 25%.
• Migrated 40+ domains from unmanaged on-premise servers to a virtualized infrastructure, automating DNS and PHP-MySQL configurations to improve maintainability and achieving a 99.99% uptime during the transition.
• Implemented DMARC, SPF, and DKIM policies to secure enterprise email infrastructure, achieving full compliance with Canadian cybersecurity standards and mitigating spoofing threats.

Inference Gateway

An 11-provider LLM gateway in Go with five routing strategies, circuit breaking, Redis response caching, and a four-stage RAG pipeline. Instrumented end to end with Prometheus, OpenTelemetry, Jaeger, and Grafana.

Persist

In-browser AI memory manager for popular AI chat providers (ChatGPT, Claude, Gemini). Runs all-MiniLM-L6-v2 locally via Transformers.js -- semantic search, cross-platform context injection, and zero external API calls. All data stays on your device.

IAM Sentinel

AI-powered Python CLI using Claude (Anthropic) and Boto3 to analyze AWS IAM roles, produce role-level risk scores, and generate actionable least-privilege recommendations.

Vinexus

AI-assisted IDE for remote Linux environments with persistent SSH orchestration, browser-based terminal access (xterm.js, SSH2), Monaco Editor, PostgreSQL, Supabase, and AWS (EC2, ALB); deployed with Railway.

Autonomous Cloud Engineer

Agent-driven cloud automation system (Go) that analyzes AWS accounts and generates policy-validated Terraform change sets to execute approved remediations with explicit approval.

SpotMyBag

Luggage identification app using a vision model.