Open to Opportunities

Backend engineer building
scalable distributed systems

Backend Engineer (2 yrs production) + AI Researcher (1 yr MSc — Dublin City University). Built 10+ Spring Boot microservices, shipped a production LLM agent (QuerySense), and automated a 24-hour manual ETL pipeline down to under 5 minutes. Strong systems thinking across distributed architecture, query optimisation, and agentic AI.

→ Seeking Backend / AI Engineer roles — Ireland, United Kingdom, USA or across Europe

Get in Touch GitHub LinkedIn LeetCode

Years SWE

Year AI Research

10+

Microservices

98.8%

Latency Reduction

Experience

Where I've made impact

Software Engineer (Promoted from Intern)

AlignBits LLC

Mar 2023 – Aug 2024

Replaced a fragile monolith with 10 independent Spring Boot microservices using event-driven architecture (AWS SQS/SNS). Teams deployed features in parallel without coordination; release cycle dropped from 3 days → 4 hours.
Analytics dashboards were timing out at P95 12s. Added strategic composite indexes and refactored N+1 queries; latency dropped to 800ms. Dashboard errors dropped to zero.
BI team spent 3 hours daily exporting ad metrics from Meta and Instagram into CSVs manually. Built async ETL pipeline (FastAPI + Celery) syncing data every 15 minutes. Reporting latency dropped from 24 hours → under 5 minutes.
Implemented AWS health checks, auto-scaling policies, and CloudWatch dashboards. System uptime improved from 97.5% → 99.8%; on-call incidents dropped from 5/month → 1/month.
Designed event-driven backend workflows using AWS SQS/SNS; downstream services reacted autonomously to upstream state changes, reducing manual intervention to zero.
Conducted code reviews across 10 services; standardised error handling and added circuit breakers. Customer-facing errors dropped 60%; cascading failures eliminated.

Java

Spring Boot

AWS

MySQL

Python

FastAPI

Software Engineer Intern

AlignBits LLC

Jan 2023 – Feb 2023

Optimised Spring Boot REST APIs reducing response time by 20% in first 6 weeks using profiling tools — leading directly to full-time offer.
Wrote unit tests achieving 95% code coverage; caught 12 critical bugs before production.

Java

Spring Boot

MSc Thesis — Generative AI Research

Dublin City University

Sep 2024 – Oct 2025

Built and trained DCGAN and CGAN architectures in TensorFlow/Keras, generating 256×256 synthetic images for plant health monitoring datasets.
Designed novel dataset augmentation strategies achieving 15% training efficiency improvement over standard CGAN baseline.
Evaluated models against quantitative industry benchmarks across multiple metrics; results competitive with state-of-the-art baselines.

Visual dataset: coffee leaf nutrient deficiencies (K, P, N, Mn, Mg, Fe, Ca, B) and healthy conditions — used to train GANs for synthetic image generation and anomaly detection.

Python

TensorFlow

Keras

Projects

Things I've built

AI-Powered Developer Tool

QuerySense — PostgreSQL Query Optimizer

EXPLAIN ANALYZE has existed for 30 years. It prints terrifying output that most developers ignore. I built QuerySense to fix that — paste a slow query, get back plain English: what's slow, why, and exactly how to fix it. With before/after proof.

Hybrid architecture — deterministic rule engine runs first (7 rules, free, instant), Claude AI runs second for human-readable explanation and exact fix SQL
7-rule detection engine — catches sequential scans, missing indexes, unindexed JOIN columns, nested loop joins, high-cost nodes, stale statistics, and partial index opportunities
Before/after benchmarking — applies fix in a rollback transaction, measures real execution time — confirmed 98.8% latency reduction on missing-index queries
Redis caching — same query never hits the Claude API twice; history layer persists every analysis to PostgreSQL
CLI tool built with Typer + Rich — formatted terminal output with benchmark tables and evaluation results
Deployed on Railway + Neon.tech + Upstash + GitHub Pages — total infrastructure cost $0
Seeded with 261K rows across 5 tables, no indexes intentionally — so QuerySense can find and fix real slow queries

Architecture

The rule engine runs before Claude AI, not after — deterministic rules catch known patterns instantly (free, always correct). Claude only runs for explanation and nuance. Every senior engineer asks "what if the LLM is wrong?" — this answers that question.

CLI Output — Real Example

QuerySense CLI output showing 98.8% latency reduction

Real CLI output — 46.8ms → 0.032ms (↓ 98.8% faster). Plan changed from Seq Scan → Bitmap Heap Scan. Fix SQL generated and improvement confirmed at HIGH confidence.

Python

FastAPI

PostgreSQL

Redis

Docker Claude AI Rule Engine

Live Demo Source

DSA & Memory Education Tool

JavaMem — Java Memory Visualizer

Most Java memory tools show you a diagram. I found that most of them — including textbooks — quietly teach wrong mental models. So I diagnosed 8 specific misconceptions, then built a visualizer that corrects each one through how it renders, not through a tooltip or a footnote.

Each structure rendered to reflect true memory behavior — ArrayList as contiguous cells, LinkedList as scattered individual node objects, Stack as a LIFO tower
HashMap displays entries by hash bucket index with an explicit "order not guaranteed" warning
Two-phase GC — objects become GC-eligible first, collected after a random delay, reflecting non-deterministic collection
Visualizes Stack, Heap & String Pool in real time with animated reference arrows
Supports 8+ data structures with live Add / Remove controls — no re-run required
Canvas-rendered BST with dynamic layout and in-order traversal display
Simulates String interning & Integer cache — including the == trap above 127
Zero-dependency, single-file — runs in any browser

HTML

JavaScript Canvas Java Memory Model DSA

Live Demo Source

SQL & Database Education Tool

SQLMem — MySQL Query Visualizer

Most people learn SQL by memorizing syntax. I found the real problem runs deeper — learners have no mental model of what the database is actually doing when a query runs. So I built a visualizer that animates every operation: rows being scanned, filtered, joined, grouped, and mutated — live, step by step, in the order MySQL actually executes them.

Schema canvas renders CREATE TABLE statements as draggable cards with animated dashed FK arrows connecting foreign key → primary key columns in real time
JOIN animator scans rows one-by-one with a glowing beam between tables — INNER, LEFT, RIGHT, and FULL JOIN each produce visually distinct Venn diagrams and result streams
WHERE clause engine supports LIKE, IN, BETWEEN, IS NULL, IS NOT NULL — matching rows flash green, filtered rows get a strikethrough animation before removal
DML operations — INSERT rows appear with a green flash, UPDATE highlights matched rows in yellow with values applied live, DELETE strikes through then removes, TRUNCATE shows an instant wipe with a side-by-side comparison card explaining the difference
GROUP BY + aggregates — SUM, AVG, COUNT, MIN, MAX animate as proportional bar charts per group, with NULL exclusion warnings on affected groups
Execution order explainer — side-by-side view of how you write SQL vs the order MySQL actually executes it, with live step highlighting during every query run
ORDER BY, LIMIT, and OFFSET animate result slicing with stats card showing total / filtered / returned counts
Zero-dependency, single-file — runs in any browser with no build step

HTML

JavaScript

CSS MySQL SVG Animation SQL Query Engine

Live Demo Source

Cloud-Native Platform

Pexeluxe — Photography Platform

Personal cloud-based image hosting platform built for global sharing.

Built using S3 + CloudFront for global image distribution
Implemented idempotent uploads via deterministic object keys
Used S3 event triggers for automatic indexing
Designed lazy-loaded infinite scroll with layout stabilization
Enforced secure access via IAM roles & bucket policies

Lambda CloudFront

Live Demo Source

Technical Skills

Tech stack at a glance

⚙️

Backend

Spring Boot

FastAPI REST APIs Microservices OAuth2 Event-Driven

☁️

Cloud & DevOps

AWS

Docker

K8s CI/CD GitHub Actions

💻

Languages & Data

Java

Python

PostgreSQL

MySQL

Mongo

Redis

🧠

AI & LLMs

Claude API LLM Integration Prompt Engineering AI Agents

TensorFlow

Keras GANs

📊

Tools

Git

Linux

Postman CloudWatch

Education & Credentials

Academic foundation

2024 – 2025

MSc Computing (AI)

Dublin City University

Modules
Machine Learning · Data Analytics · Foundations of AI · Statistical Analysis · AI Information Seeking

2020 – 2022

Master of Computer Applications

Sinhgad Institute of Management

Modules
ML · DSA · OOP · Python · JavaScript

2017 – 2020

Bachelor of Computer Applications

Marathwada Mitra Mandal College of Commerce

Modules
Java · DBMS · Networking · Data Structures

2022

Data Structures & Algorithms

IIT Madras — NPTEL

Certified

Contact

Let's work together

Available for full-time SWE roles. Dublin-based, open to remote across Europe.

LinkedIn GitHub LeetCode Phone

prasadneeraj2898@gmail.com

Bonus — for the curious

Code Philosophy

Code should be readable

UserService.java✗ Before

// Deeply nested
public Response processUser(Request req) {
  if (req != null) {
    if (req.getUser() != null) {
      if (req.getUser().isActive()) {
        return doProcess(req);
      } else {
        throw new Ex("Inactive");
      }
    }
    throw new Ex("No user");
  }
  throw new Ex("Bad req");
}

UserService.java✓ After

// Guard clauses, flat
public Response processUser(Request req) {
  validate(req);
  User user = req.getUser();
  requireNonNull(user, "No user");
  requireActive(user);
  return doProcess(req);
}

private void requireActive(User u) {
  if (!u.isActive())
    throw new InactiveUserEx(u);
}

Interactive

The dev way

neeraj@portfolio ~ zsh

# Type 'help' to see commands

~ $ whoami

Neeraj Prasad — SWE | Dublin

2yr Backend + 1yr AI | Java, Python, AWS

✓ Seeking SWE roles

~ $

helpskillsexperienceprojectscontactsudo hire me

Backend engineer building scalable distributed systems

Where I've made impact

Things I've built

QuerySense — PostgreSQL Query Optimizer

JavaMem — Java Memory Visualizer

SQLMem — MySQL Query Visualizer

Pexeluxe — Photography Platform

Tech stack at a glance

Backend

Cloud & DevOps

Languages & Data

AI & LLMs

Tools

Academic foundation

MSc Computing (AI)

Master of Computer Applications

Bachelor of Computer Applications

Data Structures & Algorithms

Let's work together

Code should be readable

The dev way

Backend engineer building
scalable distributed systems