Producer of Bugs (Software Engineer)
Senior Backend, Data & AI Platform Engineer — Specializing in cloud-native microservices, event-driven data pipelines, serverless AI platforms, and real-time collaboration systems.

📧 sabyhahmed@outlook.com • 📍 Lahore, Pakistan
🔗 LinkedIn • Portfolio • YouTube

Lead Backend Developer with 8+ years of experience designing and shipping cloud-native microservices, serverless AI platforms, and real-time collaboration infrastructure on AWS and GCP. Specialist in Go and TypeScript with deep, production-proven expertise in:

• LLM Integration: AWS Bedrock, OpenAI, Gemini — provider-agnostic orchestration with tool/function-calling
• Vector Search: Weaviate, pgvector, hybrid semantic + keyword retrieval for AI-powered search
• Data & Analytics: AWS Athena, Glue Catalog, event-driven pipelines, analytics-ready datasets
• Multi-Agent Orchestration: Building scalable AI systems with deterministic dialog control
• Cloud Infrastructure: AWS CDK, Terraform, Kubernetes (CKA certified)

Track record of leading architecture decisions across high-complexity platforms — from AR/VR enterprise collaboration and blockchain analytics to franchise intelligence and AI-powered real estate systems.

• CKA — Certified Kubernetes Administrator
• AWS Solutions Architect — Associate
• AWS Developer — Associate
• Microsoft AZ-204 — Developer Associate
• Microsoft AZ-900 & AI-900 — Azure Fundamentals
• Azure Machine Learning — Nanodegree

Enterprise AR/VR collaboration platform for immersive remote meetings & training

• Led design and delivery of 5 production backend systems (AI agents, semantic search, real-time collaboration, SSO, serverless infra) maintaining 99.9% uptime.
• Architected a provider-agnostic LLM orchestration layer (AWS Bedrock, OpenAI, Gemini) with tool/function-calling for deterministic dialog control, serving multi-turn AI meeting bots.
• Built a multi-tenant AI document search platform ingesting PDF, DOCX, Excel, images into Weaviate with hybrid semantic + keyword retrieval.
• Engineered multi-environment serverless infrastructure using AWS CDK (TypeScript) with Lambda dependency graphs, IAM wiring, X-Ray, and CloudFormation exports.
• Implemented flexible SSO (Google, Microsoft, OAuth2, PKCE/Cognito) with JWT-based role-aware access control for enterprise multi-tenancy.
• Led AWS → GCP infrastructure migration automated via Terraform & GitHub Actions; introduced SonarQube quality gates, improving test coverage by 15% and reducing release cycle time by 60%.

Franchise intelligence platform turning KPI data into AI-powered insights

• Architected AI/LLM orchestration on AWS Bedrock (Claude 3 Haiku) to auto-generate KPI narrative summaries with async job handling via SQS.
• Delivered 7 Dockerized Lambda microservices (KPI Writer, Catalog, Watchlist, Alerts, LLM Orchestration, Scheduler) fronted by API Gateway with unified event-routing.
• Built automated Watchlist Scheduler (EventBridge cron) pulling data from Aurora PostgreSQL, generating AI snapshots, and dispatching HTML email reports with PDF attachments via SES — with SNS alerting and DLQ error handling.
• Provisioned all infrastructure with modular Terraform across dev/prod, cutting environment setup from days to minutes.

NFT Marketplace & Blockchain Analytics

• Built event-driven auction scheduler (SNS + SQS + Lambda) handling 50,000+ concurrent bids with guaranteed delivery.
• Reduced listing API P95 latency from 900 ms → 180 ms (80% improvement) via query optimisation and page-level caching.
• Architected Blockchain Eye ETL pipeline ingesting multi-chain data into Athena/Parquet data lake, enabling scalable analytics.
• Built a WebSocket gateway for real-time notifications (TypeScript + API Gateway), removing client polling and reducing server load significantly.

Go · AWS Lambda · Bedrock · OpenAI · Gemini · Redis · S3 · Polly

• Serverless Go service supporting multi-turn interview/meeting bots with strict, balanced, and freeflow conversation modes.
• Provider-agnostic LLM adapter/factory with tool/function-calling (request_clarifier, redirect_to_topic, endConversation) for deterministic, safe dialog control.
• Redis-backed session state with TTLs, pipelines, and idempotent resets — managing message history, question progress, and AI config across concurrent sessions.
• LLM-based structured note extraction and edit-note command parser; multilingual transcript support via LLM translation with JSON schema preservation.
• Integrated AWS Polly TTS (Neural → Standard fallback) for real‑time client delivery.

Python · AWS Bedrock (Titan) · Weaviate · MongoDB · Lambda · SQS · Terraform

• End-to-end AI search pipeline: ingestion of PDF, DOCX, Excel, images → MongoDB → Bedrock embeddings → Weaviate with per‑tenant isolation.
• Hybrid semantic + keyword search API combining vector similarity with tenant-scoped filters, ensuring strict data isolation while preserving retrieval quality.
• Idempotent ingestion pipeline handling INSERT/UPDATE/DELETE across multiple collections with configurable chunking, retry logic, and DLQ error handling.

AWS CDK TypeScript · Lambda · Bedrock Agents · Lex · Node.js 20.x

• Designed reusable BodeInfraStack provisioning Lambdas, IAM, logging, and tagging across dev/staging/prod with environment-specific configs (timeouts, memory, concurrency, log retention).
• Implemented Lambda factory pattern (LambdaFactory + CustomLambdaFunction) standardising Node.js 20.x provisioning — runtime, bundling, X‑Ray toggles, tagging, env injection.
• Modelled Lambdas as dependency graphs with automatic fine‑grained IAM invoke permissions and ARN injection, enabling agent-style orchestration without hard‑coded wiring.

Go · Fiber · PostgreSQL (pgx) · MongoDB · pgvector · OpenAI · Gemini · JIRA · S3

• Production Go (Fiber) REST API managing vibe/scenario lifecycle, meetings, pipelines, rooms, analytics, and real‑time SSE + WebSocket endpoints with JWT auth.
• Dual‑database architecture (PostgreSQL + MongoDB, selectable via DB_TYPE), repository pattern, rate limiting, and standard middleware stack.
• Integrated LLM providers for chat API with meeting context and vector Q&A over meetings using pgvector; built JIRA integration with tool‑style function calling for AI‑driven task management.

Terraform · GitHub Actions · ECS Fargate · App Runner · RDS Postgres · Secrets Manager · Route53 · VPC · Cloudflare

• Owned AWS infrastructure across production, beta, and nonprod — App Runner, ECS Fargate, ECR, RDS Postgres with automated snapshot DR.
• Authored GitHub Actions CI/CD pipelines with pre‑deploy gates, post‑deploy health checks, E2E smoke tests, and blue/green cutover patterns.
• Led Terraform‑only IaC migration (sunsetting CDK) with modular workspaces, one‑command environment setup, and full secrets lifecycle via Secrets Manager.
• Managed Cloudflare + Route53 DNS cutovers under live traffic; debugged VPC, security group, and secret ARN issues with postmortems within 48 hours.
• Implemented monitoring/alerting with CloudWatch alarms, Grafana dashboards, and structured logging — every customer‑facing flow alarmed and paged.

Go · Python · Aurora PostgreSQL · MongoDB · Bedrock · EventBridge · SES · Terraform

• Designed microservice architecture in Go (Fiber) and Python with repository pattern across PostgreSQL, MongoDB, and Aurora PostgreSQL.
• Built Watchlist Scheduler pipeline (EventBridge cron → Aurora → Bedrock → SES/ReportLab PDF) fully automating franchise KPI reporting with SNS alerting and DLQ-based failure handling.

• 📧 Email: sabyhahmed@outlook.com
• 🔗 LinkedIn: linkedin.com/in/sabyh
• 🌐 Portfolio: sabyh-ahmad.netlify.app
• 📺 YouTube: @SabyhAhmad

• Building AI-powered platforms with multi-agent orchestration and RAG pipelines.
• Exploring advanced LLM integration patterns and provider-agnostic abstractions.
• Contributing to open-source, creating tech content, and sharing insights on system design.
• Deepening expertise in cloud-native infrastructure and observability.

Building the future, one cloud-native microservice at a time 🚀