Aditya
Aditya Kumar Kushwaha
Building production-focused Backend & Applied AI systems — from a live SaaS product (PushPostVault) to a self-hosted PR review bot and a multi-tenant RAG engine (ArchadiLM).
PostgreSQL · Redis · Qdrant
Technical Skills
Technologies I Work With
Technologies I use professionally and the areas I'm currently exploring to build reliable backend systems.
// Production Deployments & Work
Production Systems & Applied AI Platforms
Live products and microservice ecosystems engineered for high availability, zero-data-loss storage, AST code graph intelligence, and low-latency retrieval.
PushPostVault
Live Production SaaSHigh-Concurrency Chunked Storage & Secure File Distribution Platform
A production SaaS platform engineered for large-scale file delivery and collection. Built with a Go high-concurrency API engine, resumable chunked upload pipelines with client-and-server SHA-256 verification, and an atomic Redis-backed state machine with optimistic locking on PostgreSQL. Features self-expiring download links, guest invite portals with quota limits, Cloudflare R2 object storage, and Razorpay billing.
Core Engineering Architecture
AI PR Review Bot
Live PlatformMulti-Tenant Code Review Intelligence with Neo4j Knowledge Graph & Social Engine
An enterprise-ready AI code review platform operating across three decoupled microservices (GitHub App Webhook + Node/Express API + Python/FastAPI AI Engine). It parses pull request ASTs, constructs a code dependency knowledge graph in Neo4j, traces cross-file breaking changes with grounded evidence, and allows contributors to @mention the bot directly on PR lines for interactive code discussions. Features an AES-256-GCM encrypted BYOK engine, tenant admin dashboard, and an automated engine that converts merged PRs into LinkedIn and X posts with synthesized preview cards.
Neo4j AST traversal detected that modifying InvalidateSession() directly breaks 3 downstream consumers in api/gateway/auth.go.
Core Engineering Architecture
ArchadiLM
ProductionEnterprise Multi-Tenant RAG Engine with Cross-Encoder Reranking
A high-throughput Retrieval-Augmented Generation workspace engineered to index multi-modal developer resources (PDFs, VTT transcripts, source archives, API documentation). Employs Step-Back Query Expansion with HyDE document synthesis, Qdrant HNSW vector indexing, Reciprocal Rank Fusion (RRF), Cross-Encoder reranking, and ultra-low latency SSE token streaming.
"Explain chunk integrity verification and optimistic locking"
Core Engineering Architecture
Other Explorations & Tools
Targeted prototypes and open-source modular toolkits.
AI Engineering Playground
InteractiveA modular Next.js 15 cohort workspace: one dashboard, one page per GenAI assignment — persona chat, multi-LLM voting, RAG chat, tool calling, memory agents, and MCP protocol integrations.
Verkin
In DevelopmentAn intelligent networking platform: engage in natural AI technical conversations, where topic semantics match you with other engineers discussing complementary technical problems.
// System Design & Architecture
How The Systems Are Engineered
Detailed view into the distributed state machines, Neo4j code graphs, asynchronous event workers, and multi-stage vector search pipelines.
Cloud-Native Chunked Storage Engine
PushPostVaultResumable Upload State Machine & Distributed Part Assembly
Files are chunked into content-addressable blocks with SHA-256 integrity verification. Uploads are managed by a Redis-backed atomic state machine with optimistic locking on PostgreSQL, and binary payloads are stored in Cloudflare R2 object storage with zero egress overhead.
Client Upload Request
File sliced into 10MB chunks with client SHA-256 hash pre-computation.
API Gateway & Quota
Enforces guest limits, performs atomic quota reservation on Redis keys.
Integrity & Storage
Verifies binary checksum per chunk and writes directly to object storage.
Commit & Assembly
Optimistic locking commits metadata and generates self-expiring download link.
Experience
Where I Work
Professional roles, most recent first.
Backend Developer & Applied AI — Spirehubs Softwares
Currently building an AI SaaS agent bot, and leading a major backend refactor on DWIVE, a ride-hailing platform for the Caribbean region: splitting a large FastAPI monolith into a modular structure, implementing RBAC across hundreds of routes, unifying API response formats, and migrating auth to a two-token (access + refresh) JWT system.
Full Stack Developer (MERN) — Evren Global Solutions
Shipped full-stack features end to end on the MERN stack, working across the API and the React frontend.
PROFESSIONAL JOURNEY
From IT Support to Backend Engineering
My career started in enterprise IT support and gradually evolved into software engineering. Every step has been driven by curiosity, continuous learning, and building real-world systems.
Graduate Trainee
July 2022Wipro
Started my professional career in Enterprise IT Support, working on incident management, troubleshooting, production support and system reliability.
Transition to Software Development
December 2023Self Learning
Left IT operations and started learning software development from scratch, focusing on fundamentals before frameworks.
Full Stack Developer Intern
February 2025Production Internship
Worked on multiple production applications, shipping features, fixing bugs and maintaining real-world applications deployed using cPanel.
Software Engineer
August 2025Backend Team
Converted into a full-time Software Engineer and started contributing to backend services, APIs and production systems using Node.js and FastAPI.
Building PushPostVault
PresentPersonal Engineering
Outside of work, I ship and run PushPostVault end to end — product, backend, infra and users — while learning Go, Distributed Systems and Generative AI.
Current Focus
Building for the Next Chapter
Professionally, I contribute to backend systems using Node.js, Express and FastAPI. Personally, I run PushPostVault end to end — a live send-and-collect file platform I built, deployed, and am now getting in front of real users, on Go, Cloudflare R2 and production engineering principles.
By The Numbers
Engineering Highlights
Engineering Notes
Learning in Public
Notes from building production systems, learning Go, distributed systems, AI engineering and everything in between.
Building PushPostVault in Public
A deep dive into every engineering decision, mistake and architectural trade-off while building PushPostVault from scratch.
Learning GenAI using JavaScript
Learning the Generative AI in the Javascipt from 'GenAI with Javascript 2026' Cohort from chaicode.com
Lessons from Production
Interesting backend problems I solve at work, generalized without exposing company-specific implementation.
Go Learning Notes
Concurrency, Interfaces, Memory, Context,Patterns.
Docker Learning Notes
Docker, Docker Compose, Docekr bild, RUN, Images.
Get In Touch
Interested in Backend & Applied AI engineering?
Let's talk.
I'm currently working as a Backend & Applied AI Engineer building production systems with Node.js, FastAPI and modern infrastructure. If you're hiring, collaborating on an interesting project, or just want to discuss distributed systems, I'd love to connect.