Logo
Adnan
00Intro

~/adnan $ whoami

Open to SDE 2 / Forward Deployed Engineer roles

Adnan MangaonkarSoftware Development Engineer & Forward Deployed Engineer

Architecting highly scalable, secure enterprise platforms end-to-end — from client requirements to production deployment under peak load.

View resume
Years shipping enterprise platforms
3+
Enterprise applications in production
8+
Concurrent users sustained at 0% errors
1,000+
On-time delivery leading a team of 4
100%
Goregaon, Mumbai, India · 3+ years in productionScroll, or press J

01 / About

Engineer in the room with the client, and on call for the result.

Full-stack engineer and Forward Deployed Engineer (FDE) with 3+ years of experience owning enterprise platforms from client requirements to production deployment. Spearheaded 3+ major enterprise projects as an FDE at Benchmark Computer Solutions, building and delivering Exhibition Management, CRM, CMS, Document Management, Award, and Center Management systems on React, Angular, Node.js, Next.js, .NET, MS SQL Server, PostgreSQL, Python, and Redis. Strong in system design, performance engineering, high-concurrency scaling, and project management, with AI-based project experience.

  • 300,000+attendees at the live event
  • 20,000visitors a day
  • 1.7Mrequests an hour
  • 480+requests a second
  • 25×throughput gain
  • 17,000concurrent workflows load-tested
  • 0%errors at peak traffic

02 / Impact · Exhibition Management System

From collapsing at 40 users to 1,000+ with zero errors.

  1. 01 / 05 · T-4 weeks

    It falls over at 40 users.

    I inherited the backend four weeks before a live exhibition expecting 20,000 visitors a day. Under test, 45.38% of requests failed.

  2. 02 / 05 · Audit

    2,000+ files. 22,888 anti-patterns.

    Every request opened its own connection pool. I audited the codebase, eliminated per-request pools and capped what was left.

  3. 03 / 05 · Admission control

    Give every layer a limit, and a way to say no.

    A 1,000-slot in-memory admission queue with 503 load shedding, and a layered timeout hierarchy: queue, pool, SQL, watchdog, proxy. DB sessions fell from 390+ to under 160.

  4. 04 / 05 · Fleet

    Active-active, with nothing sharing a core.

    A PM2 fleet behind IIS ARR with least-current-request balancing, CPU affinity isolating mail and CRM workers from API workers, and Redis-based cron leader election.

  5. 05 / 05 · Live event

    1,000+ concurrent users. Zero errors.

    480+ requests per second, about 1.7M requests an hour, under live peak traffic with 300,000+ attendees. Throughput up 25×.

Before · 4 weeks outAfter · live event
Concurrent users
1,000+
was 40
Error rate
0.00%
was 45.38%
Throughput
25×
was 1×
Open DB sessions
<160
was 390
1 dot = 1 concurrent user served failing

03 / Systems

How it held: three layers, each one instrumented.

No rewrite and no new hardware. The gains came from finding where requests waited, and giving every layer a limit and a way to say no.

Layered timeout hierarchy

  1. queue
  2. pool
  3. SQL
  4. watchdog
  5. proxy
  1. L1 / Database tier01 of 03

    Database tier

    390 → <160

    DB sessions

    Audited 2,000+ source files (22,888 anti-patterns) and eliminated per-request connection pools. Added capped pools, a 1,000-slot in-memory admission queue with 503 load shedding, and queue-wait vs. execution-time instrumentation.

    Layered timeout hierarchy: queue → pool → SQL → watchdog → proxy. Zero stored procedure changes.

  2. L2 / Server architecture02 of 03

    Server architecture

    Active-active

    PM2 fleet behind IIS ARR

    Least-current-request balancing with intra-cluster spillover, CPU affinity isolating mail and CRM workers from API workers on Windows Server, and Redis-based cron leader election.

    Real-time cluster telemetry dashboard: event-loop lag, pool usage, heap, queue depth.

  3. L3 / Redis and queue reliability03 of 03

    Redis and queue reliability

    0

    disconnects under load

    Root-caused recurring Redis ECONNRESET drops across idle timeouts, TCP backlog, WSL2 NAT IP drift and IPv6 dual-stack. Fixed BullMQ unresolved promises.

    Replaced a 130-line if/else dispatcher with 42 validated job handlers, plus job and SMTP timeouts.

04 / Journey

The path here, from first client to on-call owner.

  1. Education2020 — 2023

    Bachelor of Science, Information Technology

    Chetana College

    • BSc-IT Student of the Year 2023 (Overall Category), Chetana College.
  2. FreelanceApr 2022 — Aug 2022

    Freelance Web Developer

    CAP NGO

    • Designed and deployed the NGO’s web platform on the MEAN stack, increasing traffic and engagement by 25%.
  3. InternshipAug 2022 — Jan 2023

    Web Developer Intern

    WEBAPPZAP

    • Built and maintained 3 websites using .NET and SSMS, boosting user engagement by 40%.
  4. Education2023 — 2025

    Master of Computer Application

    Mumbai University

    • Published research: Trends and Threats in Biometric Data Usage: Perspective on AI-Driven Identity Recognition Systems.
  5. NowJul 2023 — Present

    Software Developer (SDE 1) & Forward Deployed Engineer

    Benchmark Computer Solutions Ltd · Andheri, Mumbai

    • Primary FDE across 3+ enterprise client lifecycles: requirements, system design, development, deployment and production support for 8+ enterprise applications.
    • Built a Unified Membership Architecture integrating multiple platforms under one identity and membership layer.
    • Led a team of 4 on an end-to-end solution with an integrated CRM: sprint planning, code reviews, architecture decisions and client communication. 100% on-time delivery, 30% higher client satisfaction.
    • Re-engineered the Exhibition Management System to sustain 1,000+ concurrent users at 480+ requests/sec with 0% errors.

  6. Next

    SDE 2 or Forward Deployed Engineer.

    Start a conversation

05 / Work

Selected work.

Client platforms I owned end to end, and products I built to learn in public.

Deep dives

06 / Toolkit

The toolkit, and how it connects.

Nine architectural clusters around one operator — from client requirements and systems engineering to high-concurrency scaling, security, and cloud deployment.

toolkit / skills graph
  • Forward Deployed Engineering (FDE)
  • Client Requirements & SRS
  • System Architecture (HLD / LLD)
  • Unified Membership & SSO
  • Legacy Strangler Fig Migration
  • Multi-Tenant Architecture
  • Sprint Leadership & Code Reviews
  • React (18 / 19)
  • Next.js (App Router / RSC)
  • TypeScript
  • JavaScript (ES6+)
  • Angular (v13–v21)
  • Tailwind CSS (v3 / v4)
  • Framer Motion & GSAP
  • Redux Toolkit & Zustand
  • TanStack Query & Axios
  • Material-UI (MUI)
  • Three.js & WebGL
  • Recharts, Chart.js & D3
  • Progressive Web Apps (PWA)
  • React Hook Form & Zod
  • Node.js & Express.js
  • .NET / ASP.NET Core & C#
  • Python (FastAPI / Flask / Django)
  • NestJS
  • REST API Architecture
  • GraphQL
  • WebSockets & SSE
  • Cron Job Schedulers
  • PDF & Excel Generation (jsPDF / xlsx)
  • Nodemailer & Resend SMTP
  • Microsoft SQL Server (T-SQL / SSMS)
  • PostgreSQL
  • Supabase (Postgres & RLS)
  • MongoDB & NoSQL
  • Redis (Caching & Locks)
  • BullMQ Queues
  • Entity Framework Core
  • Drizzle ORM & Prisma
  • Database Indexing & Query Tuning
  • Capped Connection Pools
  • 1,000+ Concurrent Users (1.7M+ Req/Hr)
  • In-Memory Admission Queuing
  • 503 Automated Load Shedding
  • Layered Timeout Hierarchy
  • Active-Active PM2 Clustering
  • IIS ARR Reverse Proxy Balancing
  • CPU Worker Affinity & Isolation
  • Redis Distributed Leader Election
  • Apache JMeter & k6 Simulations
  • Memory Leak & Event Loop Profiling
  • TCP ECONNRESET Resolution
  • JWT & Secure Cookies
  • OAuth 2.0 Authorization
  • pgcrypto & bcrypt Blowfish
  • Row Level Security (RLS)
  • OWASP Top 10 Hardening
  • Rate Limiting & Honeypots
  • Strict Security Headers (CSP / HSTS)
  • Zod Runtime Validation
  • Zero-Leak Secret Management
  • Docker & Multi-Stage Builds
  • Docker Compose Orchestration
  • Kubernetes (K8s)
  • GitHub Actions (CI/CD)
  • Windows Server & IIS Hosting
  • Linux Server (Ubuntu / systemd)
  • Nginx & Apache Reverse Proxies
  • Let’s Encrypt SSL/TLS Automation
  • Git & Monorepo Workflows
  • Vercel (Edge, Serverless & Crons)
  • Google Cloud Platform (GCP)
  • Cloudflare (DNS & Edge CDN)
  • Supabase Cloud
  • Cluster Telemetry Dashboards
  • Structured Logging (Winston / Pino)
  • Health Check Keep-Alive APIs
  • Automated Uptime Watchdogs
  • Sentry Error Tracking
  • Next.js Metadata & Dynamic Sitemaps
  • Schema.org JSON-LD Structured Data
  • Core Web Vitals Optimization
  • Google Tag Manager & GA4
  • Retrieval-Augmented Generation (RAG)
  • Vector Databases (pgvector / MiniLM)
  • Autonomous AI Agents (LangChain)
  • OpenAI, Claude & Gemini APIs
  • Deterministic Intent Routing Engines
  • In-Browser ML (@xenova/transformers)

07 / Hive mindlive data

One mind. 124 nodes. Still growing.

Every product I ship reports back to one shared graph. Each cell is a real record: a feature, a guest session, a piece of retrievable content.

124
nodes
3
colonies
19
features

Colony 01 / 03 · 54 nodes

Consulting hub

This site. The projects and the content its assistant retrieves from all live here as nodes.

  • Exhibition Management System (EMS)Flagship high-concurrency platform (Node.js, SQL Server, Redis, React) scaled for 300,000+ attendees and 1.7M req/hour.
  • Database & Concurrency EngineConnection pool elimination, 1,000-slot in-memory admission queue with 503 load shedding, and layered timeout hierarchy.
  • Server Fleet & InfrastructureActive-active PM2 fleet behind IIS ARR (least-current-request), CPU worker affinity, Redis cron election, and cluster telemetry.
  • Enterprise Client PlatformsUnified Membership Architecture, Center Management, Award System, Document Management, and Three.js CAD Floorplan.
  • Core Technical ServicesProduction software architecture, web engineering, AI workflow automation, data engineering, and system security.
  • Production Case StudiesAudited production case studies: Global Data Pipeline, Enterprise CRM, Docs Wise Insights, and PetrolRadar.
  • Academic Research & PublicationsPublished research on Biometric AI threats (IJSRET/Zenodo) and RAG architecture evaluation in legal technologies.
  • Portfolio AI & Quotation EngineLocal deterministic intent scoring engine, project configurator, MiniLM 384d vector embeddings, and automated quoting.
  • Flagship Ecosystem SuiteLive production ecosystem applications: FinXpense wealth terminal, ourCloset AI wardrobe, Brighter Side, and AntiSlop.
  • 8 × Enterprise Platform Architecture
  • 7 × Server & Fleet Telemetry
  • 6 × Database Optimization & Pools
  • 5 × Production Case Study & Proof
  • 5 × Core Consulting Service
  • 4 × Local Agent & RAG Pipeline
  • 4 × High-Concurrency Engineering
  • 4 × Ecosystem Flagship Application
  • 2 × Academic Research & Publications

Colony 02 / 03 · 29 nodes

Adnan OS

The browser desktop and terminal. Every guest session leaves a trace in the hive.

  • Terminal CLI InterfaceHigh-performance virtual terminal with command parsing, tab completion, ANSI rendering, and cyber themes.
  • AOS Autonomous AgentLocal LangChain agent executing guest automations, semantic knowledge retrieval, and terminal sessions.
  • Guest Automation & TelemetryTracks active guest sessions, intent transitions, command dwell time, and interaction heatmaps.
  • Command Execution RegistryAddressable terminal commands routing to portfolio data, live projects, contact forms, and system utilities.
  • 13 × Terminal Command Execution
  • 8 × Guest Session Telemetry
  • 4 × Local Agent & RAG Pipeline

Colony 03 / 03 · 41 nodes

SortStory

The algorithm visualizer. Algorithms and runs report back to the same graph.

  • Sorting Algorithms Shell18 sorting algorithms running on non-blocking asynchronous generators for real-time 60 FPS DOM state rendering.
  • Pathfinding & Graph ShellHeuristic graph exploration engines including A*, Dijkstra, BFS, and DFS with real-time grid telemetry.
  • Binary Trees & Hierarchy ShellBinary search tree construction, node insertions, and recursive/iterative tree traversals.
  • Linked Data Structures ShellSingly and doubly linked lists, cycle detection algorithms, middle-node finding, and pointer reversals.
  • Search & Optimization ShellBinary and linear search mechanics with step-by-step partition visualization and logarithmic boundary halving.
  • Cellular Automata & ML ShellConway's Game of Life cellular automaton matrix simulation and K-Means unsupervised iterative centroid clustering.
  • 12 × Sorting Algorithm (Comparison & Exchange)
  • 6 × Sorting Algorithm (Divide-and-Conquer)
  • 5 × Linked Data Structure Operations
  • 4 × Pathfinding & Graph Traversal
  • 4 × Binary Tree & Hierarchical Structure
  • 2 × Search & Pattern Matching
  • 2 × Cellular Automata & Clustering

It grows every time someone uses what I build.

Each app posts to one endpoint; a new record becomes a new cell, and repeat activity makes a cell grow. The full version is a 3D graph you can rotate and inspect.

Enter the hive mind

08 / Credentials

Education, research and the questions people ask.

Education

  • Master of Computer Application

    Mumbai University

    2023 — 2025

  • Bachelor of Science, Information Technology

    Chetana College

    2020 — 2023

Achievements

  • BSc-IT Student of the Year 2023 (Overall Category), Chetana College.
  • Founded the TechnoGyan, TechnoDays and Technizia tech communities and exhibitions, engaging 500+ students.

Quick answers

Who is Adnan Mangaonkar?

Adnan Shafiq Mangaonkar is a full-stack Software Development Engineer and Forward Deployed Engineer based in Mumbai, India, with 3+ years of experience owning enterprise platforms from client requirements to production deployment.

What does Adnan do as a Forward Deployed Engineer?

At Benchmark Computer Solutions he is the primary FDE across 3+ enterprise client lifecycles: gathering requirements, designing systems, building, deploying and supporting 8+ enterprise applications including Exhibition Management, CRM, CMS, Document Management, Award and Center Management systems.

What is his most significant engineering result?

He re-engineered an Exhibition Management System that collapsed at 40 concurrent users so that it sustained 1,000+ concurrent users at 480+ requests per second (about 1.7M requests per hour) with 0% errors, four weeks before a live event with 300,000+ attendees.

Which technologies does he work with?

Full-stack & FDE engineering across React 19, Next.js 16, Angular, Node.js, .NET Core & Entity Framework, and Python; MS SQL Server, PostgreSQL, Supabase, MongoDB, Redis and BullMQ; PM2 active-active clustering, IIS ARR, Docker, Kubernetes, GCP, Vercel and Cloudflare; plus load simulation (JMeter), cybersecurity (OWASP, RLS, pgcrypto), and AI workflows (RAG, LangChain, pgvector).

What roles is he open to?

SDE 2, Forward Deployed Engineer and senior full-stack roles, based in Mumbai and open to remote work with global teams.

09 / Contact

Need someone who owns it through production?

I am open to SDE 2 and Forward Deployed Engineer roles, from Mumbai or remote. The resume has the detail; a call is faster.

Get in touch Book a call Resume PDF

❯ try , or in the command palette, or press Ctrl+` for the terminal