About Projects Experience Contact

Hi, I'm

NAYEEM
FARDIN

Systems Engineer

Scroll
AWS KUBERNETES FASTAPI REACT KAFKA RABBITMQ TERRAFORM POSTGRESQL GO AWS KUBERNETES FASTAPI REACT KAFKA RABBITMQ TERRAFORM POSTGRESQL GO

ABOUT.

I'm a Computer Science undergrad at Islamic University of Technology who builds things that scale and break gracefully. My sweet spot is the intersection of ML systems, cloud infrastructure, and backend engineering — I enjoy taking models from notebooks to production with proper observability, CI/CD, and cost-awareness.

Education

B.Sc. in Computer Science & Engineering

Islamic University of Technology (IUT)

Expected Sep 2026

IUT Logo

Tech Stack

Languages
Python Go JavaScript C++ Java
Web & API
React Node Express FastAPI
AI & ML
LangChain PyTorch RAG OpenAI
Cloud & Ops
AWS Docker Kubernetes Terraform Linux Git
Sys & Tools
Kafka RabbitMQ Celery Grafana ArgoCD Helm
Databases
PostgreSQL Redis InfluxDB Pinecone MongoDB

PROJECTS.

Python YOLOv11 AWS EKS FastAPI Docker ONNX Prometheus ArgoCD Helm
  • Fine-tuned and deployed YOLOv11 on AWS EKS for dense retail product detection.
  • Scalable inference service using Helm + ArgoCD (GitOps-driven deployments).
  • CPU inference speedup and 70–90% cost savings via INT8 quantization, AMX and Spot diversification.
  • Canary releases with Argo Rollouts + smoke tests; Loki/Prometheus/Grafana for logging and observability.
  • DevSecOps in GitHub Actions with Trivy and pip-audit for secure ECR promotion.
Go Python Celery RabbitMQ Terraform AWS K3s Redis
  • Migrated the HTTP edge from FastAPI to Go, preserving full contract parity with the existing Celery worker runtime and Alembic-managed schema.
  • Architected a dual-node K3s cluster on AWS (On-Demand control + Spot data plane) with full Terraform IaC and automated SSM reconciliation for async image processing queued through RabbitMQ.
  • Optimized cost with EC2 Spot and graceful eviction via AWS Node Termination Handler.
  • Resilience under load with Redis idempotency keys, pybreaker circuit breakers, and k6 stress tests.
  • Zero-Trust GitHub Actions (OIDC) CI/CD; telemetry and traces to Grafana LGTM via Alloy.
  • Decoupled ML inference from fast conversions via dedicated queues to prevent worker starvation.

StatusMonitor

FastAPI React 19 Apache Kafka InfluxDB PostgreSQL Docker
  • Distributed monitoring platform with FastAPI microservices (ingestion/auth/alerting) across Docker, designed for independent scaling.
  • Windows agent (psutil + PDH) streaming metrics via Kafka to decouple producers/consumers.
  • Reduced alert noise with in-memory cooldowns; real-time Telegram alerts + full history for auditability.
  • Lowered storage cost using InfluxDB downsampling + tiered retention; secured onboarding with Argon2 + JWT.

Crawler RAG DocHelper

Python LangChain Pinecone Gemini 2.5 Cohere
  • Hybrid-search RAG for docs using Semantic (HuggingFace) + Keyword (BM25) retrieval in Pinecone to boost relevance.
  • Automated ingestion by crawling llms.txt via Tavily with caching to avoid redundant fetches.
  • Added Cohere Rerank before Gemini 2.5 generation to reduce noisy context.
  • Exposed query-time alpha-weighting to tune keyword precision vs semantic recall.

interviewPrepper

LangGraph FastAPI React Firecrawl Multi-LLM
  • Orchestrated multi-step AI agents via LangGraph and LangChain to automate 30-day curriculum generation.
  • Integrated cross-LLM routing natively supporting OpenAI, Google Gemini, and OpenRouter architectures.
  • Implemented active AI interaction with persistent SQLite conversation threading allowing user refinements.
  • Built robust external data retrieval merging Firecrawl web-scraping with Tavily AI search fallbacks.

RoutineMaker

FastAPI React Docker PostgreSQL Nginx
  • Microservices architecture with distinct Auth and Core Backend FastAPI instances isolated via Docker.
  • Secured endpoints using Argon2 hashing, JWT authentication, and pre-configured CORS protection.
  • Abstracted multi-service routing cleanly behind an Nginx API Gateway container.
  • Automated PDF generation of custom schedules for local export.

EXPERIENCE.

Backdoor Private Limited Logo

CyberSecurity Internship

Backdoor Private Limited

October 2025
  • Digital forensics analysis to recover sensitive deleted artifacts using FTK Imager and Autopsy.
  • Security assessments on Android applications via MobSF and web platforms using Burp Suite and SQLMap.

LET'S BUILD
SOMETHING.

Got a project idea, internship opportunity, or just want to chat? Hit me up.