ARCHITECTURE
Systems-First Architecture
Establish strict database schemas, API contracts, and failure boundaries prior to client execution.
Rajendra
Bist
>Full-Stack Developer & Backend Engineer
Backend-focused full-stack developer (Node.js, TypeScript, PostgreSQL) designing and shipping complete applications end to end: schema design, REST APIs, authentication, tenant isolation, and deployment. Open to roles in Delhi NCR and remote.
Rajendra BistONLINE [UTC+5:45]
01 // EXPLORE SYSTEMS
01 // ARCHITECTURE & PHILOSOPHY
Engineering Systems with Rigor
Representing engineering identity through structured schemas, strict typing, and fault-tolerant pipelines.
rajendra.config.ts
I'm a full-stack developer and backend engineer based in New Delhi, India working primarily with Node.js, TypeScript, PostgreSQL, Express, and Next.js 15. Originally from Nepal, I am 100% legally authorized to live and work across India without visa sponsorship or a work permit under the 1950 Indo-Nepal Treaty of Peace and Friendship, and am actively available for roles in Delhi NCR or remote.
My engineering strategy is systems-first: define rigid database models, enforce schema validation, and architect fallback paths before writing UI code. I've shipped production systems including AST static analysis review agents, vector similarity search with pgvector, and resilient multi-model LLM streaming pipelines.
In AI engineering, I pair probabilistic models with software controls: grounded retrieval, schema guardrails, bounded outputs, and explicit fallback behavior that makes failure visible and testable.
TYPESAFE
Contract-Driven Full-Stack
Bridge complex client state with typed backend endpoints and make data-shape failures easier to detect.
PERFORMANCE
High-Throughput Backend
Architecting secure authentications, rate limiting, MongoDB & PostgreSQL transactions, and caching.
RELIABILITY
Deterministic AI Guardrails
Grounded retrieval pipelines and multi-tier LLM fallback routers that bound failure and preserve context.
02 // TECH ARSENAL
Production Stack & Tooling
Technologies verified through real-world deployment, typed contracts, and measurable throughput.
Node.js
CORETypeScript
COREExpress.js
COREREST APIs
COREZod Validation
COREJWT Security
COREWebSockets
COREPostgreSQL
COREMongoDB Atlas
CORESupabase pgvector
COREMongoose ORM
COREUpstash Redis
COREDocker
COREGit & GitHub
COREGitHub Actions CI/CD
COREVercel Edge
CORELinux / Bash
CORENext.js 15 (App Router)
COREReact 18.3
CORETailwind CSS v4
COREFramer Motion
CORERAG Pipelines
COREGoogle Gemini 2.5
COREGroq LLM Engine
CORETool-Calling Agents
COREVector Embeddings
CORELangChain.js
EXPLORINGLLM-as-Judge Evals
EXPLORING03 // FEATURED PROJECTS
Production Systems & Case Studies
Real-world architectures structured around Problem, Solution, Key Decision, and Frontend Implementation.
IN DEVELOPMENT • 2026
Sovereign
Multi-Tenant B2B SaaS for Autonomous AI Workflow Agents
The Problem
“Multi-tenant AI SaaS platforms often rely on leaky application-level tenant filters and fragile, stateless LLM executions that fail silently or leak confidential data across organizational boundaries.”
The Solution
“Architecting a backend-heavy, multi-tenant platform in a Turborepo monorepo with PostgreSQL Row-Level Security, Inngest durable workflows, human-in-the-loop approvals, and schema-guarded tool loops.”
Key Architectural Decision
Enforced tenant isolation at the database layer via PostgreSQL Row-Level Security (RLS) and orchestrated durable, resumable step executions with Inngest.
Frontend Implementation
Architected organization workspace in a Turborepo monorepo featuring human-in-the-loop agent review queues, audit logs, and tenant isolation dashboards.
Postgres RLS
DB Tenant Isolation
Inngest
Durable Runs
Zod-Guarded
Agent Tool Loop
TypeScriptTurborepoPostgreSQL (RLS)InngestZodNode.jsNext.js 15
PRODUCTION SYSTEM • 2026
DevGuard AI
Autonomous PR Security & Code Review Agent
The Problem
“Most AI PR reviewers hallucinate syntax and miss security vulnerabilities because they lack runtime static analysis and test validation.”
The Solution
“Empirical tool-calling review loop that executes AST linters and vulnerability scanners before synthesizing 1-click GitHub PR review comments.”
Key Architectural Decision
Engineered a three-path model router (Groq Llama 3.3, Gemini 2.5 Flash, and deterministic AST rules) so provider failures have an explicit fallback path.
Frontend Implementation
Designed the dashboard experience around traceable review states, exposing agent steps, PR simulation, and review evidence without hiding the backend verification flow.
3-Tier
Failover Router
5 Max
Agentic Loop Cap
AST-Verified
Syntax Assurance
Next.js 15React 18.3TypeScriptTailwind v4SupabaseGroq 70BGemini 2.5GitHub App API
PRODUCTION SYSTEM • 2026
EduMethod AI
Cognitive EdTech Platform With Persistent Memory & pgvector RAG
The Problem
“Generic AI chatbots forget student weak areas between sessions and hallucinate answers on complex syllabus materials without grounding.”
The Solution
“End-to-end cognitive workspace pairing long-term student memory profiles in Supabase with pgvector document grounding and an independent verification audit layer.”
Key Architectural Decision
Chose self-hosted pgvector inside PostgreSQL over external vector databases to eliminate cold starts and keep user data and vector embeddings in a single atomic transaction.
Frontend Implementation
Built the learning workspace around authenticated Next.js/React flows that connect upload, memory, verification, and spaced-repetition states to typed backend responses.
pgvector
Syllabus Grounding
SM-2
Spaced Repetition
7 Engines
Cognitive Core
Next.js 15React 18.3TypeScriptClerk AuthSupabase (pgvector)Upstash RedisGroq RouterGemini 2.5
PRODUCTION SYSTEM • 2026
Chronicle
Publishing Platform With Verified Accounts & Schema-Guarded AI Writing Tools
The Problem
“Publishing apps with AI features usually trust model output blindly and skip the operational basics: verified accounts, transactional email, crawlable rendering, and migration-safe deploys.”
The Solution
“Two-service platform (Next.js on Vercel, Express/Prisma/PostgreSQL on Render) with server-rendered public pages, email-verified accounts, and Gemini tools whose structured output is schema-validated before it reaches the editor.”
Key Architectural Decision
Treated the LLM as an untrusted dependency: JSON-mode requests, Zod-validated output, normalized tags, hard timeouts, and a single uniform failure path so provider errors never leak to users.
Frontend Implementation
Server-rendered public reading pages with ISR, per-page metadata and dynamic social cards, plus a client-side writer workspace with markdown preview, draft autosave, and AI assist panels on a tokenized dark/light design system.
Zod-Guarded
AI Output Contract
SSR + ISR
Public Reading Surface
CI-Gated
Tests on Every Push
Next.js 15TypeScriptExpressPrismaPostgreSQLGemini 2.5Brevo EmailVitestGitHub Actions
04 // ENGINEERING PROCESS
How I Architect & Ship Systems
A four-stage delivery process that makes assumptions explicit, validates behavior, and keeps failure modes visible.
01CONSTRAINTS & SCOPE
Understand
Map data flows, failure modes, rate limits, and latency budgets before writing code.
02SCHEMA & CONTRACTS
Design
Model database schemas, establish strict JSON schema contracts, and architect fallback routers.
03TYPESAFE EXECUTION
Build
Develop vertical slices with full TypeScript type-safety, robust validation, and clean state handling.
04CI/CD & MONITORING
Deploy
Continuous deployment with edge caching, automated build validation, and uptime telemetry.
05 // EXPERIENCE & TIMELINE
Work & Systems Timeline
Track record of shipping production APIs, AI workflows, and type-safe web applications.
Independent Projects
Full-Stack Developer (Self-Employed)
Feb 2025 – PresentRemote
Design, build, and deploy full-stack applications end to end across Next.js (Vercel) and Express (Render) with PostgreSQL (Prisma) and MongoDB, owning schema design, REST APIs, authentication, CI, and production deployment.
Architecting Sovereign: multi-tenant B2B SaaS in a Turborepo monorepo where organizations define autonomous AI workflow agents; PostgreSQL Row-Level Security enforces tenant isolation at the database layer alongside RBAC, audit logging, and quota enforcement.
Built workflow execution on Inngest for durable, resumable, and idempotent runs with human-in-the-loop approval steps, and an agent tool-calling loop treating the LLM as an untrusted dependency via JSON-mode requests, Zod-validated outputs, hard timeouts, and prompt-injection tag defenses.
Built and deployed Chronicle: two-service platform with Next.js SSR/ISR on Vercel and Express REST API (JWT) on Render backed by PostgreSQL via Prisma; schema-guarded Gemini tools with transactional email (Brevo) and boot-time zero-drift Prisma migrations.
Engineered DevGuard AI: GitHub App with LLM tool-calling loop (AST-based security linting & OSV.dev CVE scanning), 3-path model failover (Groq → Gemini → rule engine), HMAC-verified webhooks, Zod validation, and Redis rate limiting.
TypeScriptTurborepoPostgreSQL (RLS)InngestNode.jsExpressPrismaNext.js 15ZodGemini 2.5VitestGitHub Actions
Vcare Technical Institute
Full-Stack Web Development Trainee
2025 – 2026India
Completed practical diploma training in full-stack engineering focused on API engineering, database schema modeling, protected authentication flows, and deployment patterns in India.
Implemented REST API modules, protected JWT authentication routes, database schemas, and reusable React interface patterns.
Practiced production workflows including Git, environment variables, CI deployment checks, and debugging discipline.
Node.jsExpress.jsReactMongoDBREST APIsGit
06 // CREDENTIALS & TRAINING
Education & Verified Training
Formal software engineering foundation and industry-standard technical accreditations.
Completed
Professional Engineering Training
Diploma in IT (Full-Stack Web Development)
Vcare Technical Institute
2025 – 2026
Rigorous practical engineering diploma covering Node.js, Express.js, React, MongoDB, REST APIs, database modeling, and production deployment patterns in India.
Core Focus & Topics
Node.jsExpress.jsReactMongoDBREST APIsPostgreSQL
Completed
Class XII Science Equivalent
Higher Secondary (+2 Science)
Kailali Multiple Campus
2022
Formal academic foundation in physical sciences, advanced calculus, physics, and computational reasoning in Nepal.
Core Focus & Topics
MathematicsPhysicsScientific ComputingLogic & Analysis
Ongoing
Applied Self-Directed Engineering
Continuous Systems Engineering
Project-Based Engineering Practice
Ongoing
Architecting and shipping production-focused platforms including Sovereign (Multi-Tenant B2B SaaS on PostgreSQL RLS), Chronicle (two-service publishing platform), and DevGuard AI (autonomous GitHub code review agent).
Core Focus & Topics
PostgreSQL RLSTurborepo MonorepoInngest WorkflowsZod GuardsAST Security Analysis
07 // CONTACT & INQUIRIES
Initiate a Collaboration
For full-stack, distributed backend, and AI infrastructure work, send a concrete problem or role brief.