Projects Infrastructure Skills Experience Contact GitHub → Download Resume
Open to backend engineering roles

Mohammed
Yusuf

Backend Engineer · 3 years at TCS building Java microservices and Kafka event pipelines at enterprise scale. Self-hosted infrastructure across a home server and Oracle Cloud — 20+ Dockerized services, zero-trust networking, full observability — plus a private agentic AI system I built from scratch (local LLMs, RAG, tool-calling).

// Systems that stay up at 3 am  ·  reliability over ceremony
Download Resume View GitHub Get in Touch
3+
Years at TCS
~1.2k
msg/sec Kafka
22
Docker Services
P99 550ms
Latency (was 850)
Engineering Work

Projects

Enterprise-scale work at TCS and production systems I own end-to-end.

01 TCS · 2023

Event-Driven Kafka Transaction Pipeline

Problem: A 3-hop synchronous REST chain processed each transaction sequentially — any single-hop failure broke the whole chain, and month-end batch load pushed P99 latency to 850ms with routine consumer timeouts.

Replaced the blocking chain with an async Kafka pipeline — partitioned consumers, a dead-letter queue with retry handling, and idempotent settlement writes. P99 latency dropped from 850ms to 550ms at ~1,200 msg/sec peak, scaling horizontally by adding consumers with no code changes.

550msP99 Latency
~1.2kmsg/sec Peak
6Partitions
DLQ3× Retry
Apache Kafka 3.x Spring Boot Java DLQ Idempotent Consumer Exp. Backoff
02 TCS · 2022

Legacy Mainframe Migration & CI/CD Pipeline

Problem: No test coverage, 14-minute manual build, and frequent critical issues shipping to production.

Migrated .NET and mainframe systems to Java Spring Boot microservices — 40+ REST endpoints, 35% faster execution, high unit test coverage enforced at every release. Built Jenkins CI/CD pipelines with staged quality gates and auto-rollback; reduced production defects by 30%.

40+Endpoints
11minBuild Time
30%Defects ↓
0Critical Issues
Java Spring Boot Docker Jenkins JUnit SonarQube Auto-rollback
03 Agentic AI · Self-Built

Falak AI — Agentic AI Execution System

Design goal: A private, self-hosted agent that plans, calls tools, and executes — minimizing cost while keeping data on my own hardware.

Built from scratch: an agentic loop with planning, tool-calling, and ReAct-style execution over 8 tools (shell, file, web search, Notion, n8n). Smart model routing scores query complexity to send simple work to a local LLM (Ollama) and hard problems to the Anthropic API. A 3-tier memory system (SQLite sessions + ChromaDB semantic recall + typed memory files) adds RAG ingestion, auto-memory extraction, and context compaction. Shipped with a streaming Telegram interface, FastAPI REST, and CLI — guarded by prompt-injection detection, an audit log, allowlist access control, and human-in-the-loop approval for destructive actions.

8Agent Tools
3-TierMemory + RAG
Local+CloudModel Routing
HITLApproval Gate
Python FastAPI Ollama Anthropic API ChromaDB ReAct Agent Docker
github.com/md-yusuf-f/falak-ai
04 Async · Finance

Falak Finance — Real-Time Portfolio Monitor

Design goal: Aggregate 3 brokerage APIs into a single observable dashboard with parallel AI analysis and automated alerting.

Real-time portfolio dashboard aggregating equity, crypto, and mutual fund data from 3 brokerages. Five AI models analyse holdings in parallel with graceful degradation; automated alerting, paper trading engine, and scheduled reporting.

5Parallel Analysts
3Data Sources
5kCandle BT
15minSignal Cycle
Python FastAPI asyncio ccxt / Binance KiteConnect APScheduler Docker Circuit Breaker
github.com/md-yusuf-f/falak-finance-monitor
05 Infrastructure · Open Source

Markaz — Self-Hosted Ops Dashboard

Design goal: Single-pane real-time visibility into a Docker homelab — containers, metrics, services, and network — with zero cloud dependency.

Real-time ops dashboard for Docker-based homelabs built with React + Vite. 15+ live data panels pulling from Prometheus, Docker socket-proxy, Glances, Uptime Kuma, Traefik API, and Alertmanager. Container restart actions, multi-node health bar with per-node CPU/RAM/Disk, live network RX/TX graph, and Traefik route inspection — all in an eDEX-inspired full-screen interface.

15+Live Panels
6Data Sources
ARM64Compatible
0Cloud Deps
React Vite Nginx Prometheus Docker Glances Uptime Kuma Traefik API
github.com/md-yusuf-f/markaz
Self-Hosted Systems

Infrastructure &
Operational Ownership

20+ Dockerized services across a home server and an Oracle Cloud ARM64 VPS — encrypted offsite backups, SSO-gated access, and full Prometheus observability. Traefik for ingress, CrowdSec for threat intelligence, Tailscale mesh and Cloudflare Tunnels for zero-trust access.

Real reliability engineering: capacity planning, incident response, and on-call at 3am — no managed services, no safety net.

Traffic & Entry
Cloudflare CDNDDoS protection, zero open ports
Cloudflare TunnelEncrypted inbound, no firewall rules
Traefik v3TLS termination, middleware routing
AutheliaSSO + TOTP 2FA on all interfaces
Hosts & Topology
Home Server
Primary compute
Bulk of services + data tier
Oracle Cloud
ARM64 VPS
Public edge services
Link
Tailscale mesh
Private zero-trust network
Core Services
VaultwardenPassword manager
NextcloudFile sync & storage
Falak AILocal LLM assistant
n8nWorkflow automation
+ 16 more services
Observability
PrometheusHost, container & JVM metrics
GrafanaDashboards & visualization
AlertmanagerSLO alerts → Telegram
Uptime KumaExternal endpoint monitoring
Security & Backup
CrowdSecCommunity threat intelligence
TailscalePrivate management VPN
rcloneAES-256 encrypted nightly backup
Retention7 daily · 4 weekly snapshots
Core Capabilities

Technical Skills

Full ownership from system design to production observability.

Backend Engineering

P99 850ms → 550ms on Kafka pipeline

Event-driven microservices, REST APIs, and async processing with Java and Spring Boot at enterprise scale.

  • Spring Boot and Hibernate ORM at enterprise scale
  • Apache Kafka — partitioned topics, dead-letter queues, at-least-once delivery
  • Idempotent consumers with exponential backoff retry
  • Legacy .NET and mainframe migration to Java microservices

Cloud & DevOps

20+ Docker services · zero-trust mesh · offsite backups

CI/CD pipelines, container orchestration, and Linux server management across a home server and Oracle Cloud ARM64.

  • Docker and Jenkins CI/CD with staged quality gates and auto-rollback
  • Traefik v3 with automated TLS via Let’s Encrypt
  • Tailscale mesh VPN with UFW default-deny firewall
  • AES-256 encrypted offsite backups via rclone

Quality & Observability

High coverage · 0 critical SonarQube issues enforced

High-reliability pipelines with full-stack observability and active threat detection in production.

  • JUnit and Mockito with high coverage gates enforced in CI
  • SonarQube quality gates blocking releases on critical issues
  • Prometheus, Grafana, and Alertmanager for production observability
  • CrowdSec threat detection with Authelia SSO and 2FA
Specialized Engineering

Agentic AI & Automation

8-tool agent · local+cloud routing · RAG memory

Private, self-hosted agentic AI with planning, tool-calling, and complexity-based model routing.

  • Agentic loop with planning, tool-calling, and ReAct-style execution
  • Smart model routing — local Ollama vs Anthropic API by query complexity
  • 3-tier memory (SQLite + ChromaDB) with RAG ingestion and auto-extraction
  • n8n workflow automation and Telegram bot integrations

Azure Security

Microsoft Certified · 2023–2027

Cloud security architecture, identity management, and compliance posture on Azure — applied to OCI infrastructure hardening.

  • Azure IAM & access control
  • Defender for Cloud monitoring
  • Network security groups & private endpoints
  • Key Vault & encryption practices
About

Production at work.
Production at home.

Backend Engineer at Tata Consultancy Services — 3 years building Java / Spring Boot microservices at enterprise scale. Kafka event pipelines, legacy migrations, CI/CD quality gates, high test coverage enforced on every release.

Outside work: Falak — self-hosted infrastructure across a home server and an Oracle Cloud ARM64 VPS, 20+ services, zero-trust networking, encrypted backups, full observability. Real production engineering, no managed services.

Microsoft Certified: Azure Security Engineer Associate Credential ID: C2BED8FD1B1373F  ·  Valid 2023–2027
Daily Use
Java Spring Boot Apache Kafka Docker JUnit / Mockito PostgreSQL Jenkins Maven Python FastAPI Ollama / LLMs ChromaDB
Infrastructure
Traefik v3 Cloudflare Tunnels Tailscale UFW / Linux OCI / ARM64 Docker Compose Authelia 2FA CrowdSec
Observability
Prometheus Grafana Alertmanager SonarQube Uptime Kuma
Familiar
n8n Azure Security SQLite Telegram Bot API
Experience

Work History

Aug 2022 — Present
System Engineer
Tata Consultancy Services  ·  Bengaluru, India
  • Migrated .NET and mainframe systems to Java and Spring Boot — 40+ REST endpoints delivered, 35% faster execution, zero data loss during cutover.
  • Designed Apache Kafka event-driven pipeline — P99 latency 850ms → 550ms, 1,200 msg/sec peak; dead-letter queue with exponential backoff for consumer resilience.
  • Built Jenkins CI/CD with staged quality gates — build time 14→11 min, zero critical issues enforced on every release.
  • Maintained high unit test coverage with JUnit and Mockito — 30% fewer production defects across 6 consecutive releases.
  • Mentored junior engineers on Java and Spring Boot patterns — improved team delivery cadence by 20%.
Open Source & Code

GitHub

Infrastructure, AI, and automation projects are open. TCS work is under NDA. Architecture decisions and deployment configs documented in each README.

falak-ai

Self-hosted agentic AI — ReAct loop over 8 tools, complexity-based local/cloud model routing, 3-tier memory with RAG, Telegram/REST/CLI with human-in-the-loop approval.

Python FastAPI · Ollama · Docker
falak-finance-monitor

Real-time portfolio aggregation with 5 parallel AI analysts, paper trading engine, RSI/EMA signals, 5k-candle backtester, and Telegram alerting.

Python FastAPI · asyncio · ccxt
markaz

Real-time ops dashboard for Docker homelabs — 15+ live panels from Prometheus, Glances, Uptime Kuma & Traefik, container restart actions, and an eDEX-inspired full-screen UI.

JavaScript React · Vite · Prometheus

github.com/md-yusuf-f  ·  Architecture docs in each repository README

View All Repositories →
Education

Academic Background

B.E. in Mechanical Engineering

AVS Engineering College  ·  Tamil Nadu, India  ·  2016–2020  ·  CGPA: 7.49

// Self-taught into software — Java, Spring Boot, distributed systems, infrastructure

Contact

Get in Touch

Open to backend and platform engineering roles — distributed systems, cloud-native, or AI-augmented backend. Remote or Bengaluru.

LinkedIn GitHub Email

Mohammed Yusuf — Backend Engineer · Salem, TN → Bengaluru
Specialising in distributed systems, Kafka pipelines, cloud-native infrastructure, and agentic AI systems.

Resume