MOMIN SHAIKH

New talent

@thisismominshaikh

Senior Software Engineer

New talentFrom $75/hrBangladeshHires and manages projectsProvides services0 views

Message

Track record

Projects done

New

Talent rating

Open

Availability

About

Full-stack software engineer with hands-on experience building production-grade web applications using React, Next.js, Express.js, Django, Laravel, and AI integrations. Proven ability to design scalable architectures including multi-step agentic AI systems, multi-tenant SaaS platforms, RAG-powered AI platforms, and real-time applications. Strong foundation in data structures, algorithms, and system design. Fluent in English and Bengali.

Experience

Open-Source Developer

Independent · Remote

2020 – Present

Designed and shipped multiple full-stack open-source projects spanning agentic AI systems, SaaS platforms, real-time collaboration tools, and multi-tenant architectures - entirely self-directed. - Built CodeFusion Research Agent - a Django/Next.js AI agent using Google Gemini 2.0 Flash function calling to autonomously explore GitHub repositories and answer deep technical questions. - Engineered Collaborative Team Hub with dual-database architecture (PostgreSQL + MongoDB), real-time Socket.IO, JWT auth, RBAC, and Kanban workflows deployed on Railway. - Contributed reusable, well-documented codebases to GitHub; maintained production deployments on Vercel and Railway with CI/CD pipelines.

WebFlow Developer

Flow Seek · Khulna, Bangladesh

2020 – 2023

Education

Faculty of Law, Islamic University, Kushtia-Jhenaidah

LL.B (Hons) in Al-Fiqh and Law

2019 – 2023

CGPA: 3.28 / 4.00

Nachole Govt. College, Rajshahi Board

Higher Secondary Certificate (H.S.C), Science

2018 – 2018

GPA: 4.58 / 5.00

Maktapur High School, Rajshahi Board

Secondary School Certificate (S.S.C), Science

2015 – 2015

MOMIN SHAIKH

Discuss your project

Message

Portfolio

VentureForge AI

Next.js 15, Express.js, Node.js, MongoDB, Mongoose, Groq SDK (Llama 3.3 70B & 3.2 Vision), Better Auth, TanStack Query, Recharts, Tailwind CSS v4, Vercel | 2026 - Architected a full-stack AI startup incubator & financial validation portal connecting a Next.js 15 App Router frontend with an Express/MongoDB backend, deployed cross-origin on Vercel with Better Auth session management. - Engineered an agentic startup evaluation engine using Groq LLM (llama-3.3-70b-versatile) to auto-score venture viability, generate structured SWOT analysis, recommend tech stack components, and perform competitor gap analysis. - Built an interactive financial forecasting dashboard with CFO AI audit, ingesting CSV/manual financial ledgers to construct dynamic Recharts (Area charts for cash flow, Bar charts for net margins) and automated executive financial summary reports. - Designed VentureMentor AI chat drawer, a conversational coaching drawer grounded in venture details that maintains session history in MongoDB and generates real-time follow-up prompt suggestions. - Integrated multimodal pitch & image intelligence, using Llama 3.2 11B Vision for base64 image layout audits and document text parsing for instant pitch deck digests and risk summaries.

View project

CodeFusion Research Agent

Django, Django REST Framework, PostgreSQL, Google Gemini 2.0 Flash, Next.js 16, React 19, TypeScript, Tailwind CSS, Gunicorn, Railway, Vercel | 2026 - Architected a full-stack AI research agent that answers deep technical questions about any GitHub repository by autonomously cloning the codebase and exploring it through a bounded multi-step tool-calling loop powered by Google Gemini 2.0 Flash's native function calling - eliminating hallucination by grounding every answer in actual source files. - Designed a custom agentic execution engine (gemini_runner.py) with a hard-capped 24-round tool loop using Gemini's FunctionDeclaration + FunctionResponse pattern; dispatches 7 tools (list_files, read_file, search_code, get_file_summary, save_finding, get_previous_findings, list_past_sessions), feeds results back as structured parts, and terminates cleanly on text-only response or marks session FAILED on round cap or exception. - Engineered a 4-table PostgreSQL schema: Finding rows persisted incrementally so knowledge survives across sessions; references stored as denormalized JSONField to avoid a join hot-path on every session read; ToolCallLog written before results are fed back to the model, ensuring a crash-safe, pre-execution audit trail. - Implemented context window management with configurable hard caps - 120 KB per file read, 40 max search results, 2,000-char per-finding truncation in LLM context - and a get_file_summary cheap-outline tool to avoid full reads, bounding token costs on large codebases. - Built prior-knowledge surfacing: at session start the agent pre-loads past findings and session summaries into the initial prompt, saving 1-2 tool round-trips on repeat queries and enabling the model to build accumulated knowledge across independent sessions. - Deployed Django backend on Railway (Gunicorn + psycopg2) with environment-driven SQLite to PostgreSQL switching, CORS-locked origins, and DRF JSON-only renderer; Next.js 16 / React 19 frontend on Vercel features an AgenticLoader visualising live tool-call progress and a full ToolCallLog audit view.

View project

Collaborative Team Hub

View project

Bookified

Next.js, LangChain, OpenAI, Stripe, Clerk, Vercel AI SDK | 2026 - Engineered a production SaaS platform enabling users to chat with uploaded PDFs using a Retrieval-Augmented Generation (RAG) pipeline built with LangChain and OpenAI, delivering contextually accurate answers in real time. - Integrated bi-directional voice features via OpenAI Whisper (speech-to-text) and OpenAI TTS, enabling a fully conversational hands-free document interaction experience. - Implemented real-time response streaming with the Vercel AI SDK, delivering word-by-word AI output to significantly reduce perceived latency and boost user engagement. - Built a full SaaS billing and access-control system using Stripe (tiered subscriptions) and Clerk (auth), with custom middleware gating premium features such as voice and larger file limits behind a Pro plan. - Developed a scalable document processing pipeline with Uploadthing and PDF-parse for secure cloud upload, text extraction, and vector indexing for AI consumption.

View project

Multi-School Management System

Laravel, React, TypeScript, Vite | 2026 - Architected a multi-tenant database system supporting isolated data environments for concurrent schools using custom Laravel middleware and global query scopes, ensuring strict data privacy per tenant. - Implemented a granular RBAC system with Laravel Gates and Policies covering four roles (Admin, Teacher, Parent, Student), with dynamic UI rendering based on access level. - Engineered a dynamic GPA calculation engine adapting to diverse school-specific grading scales with automated PDF report card generation, translating complex business logic into efficient backend code. - Optimized RESTful APIs using Laravel API Resources and resolved N+1 query problems via eager loading, significantly reducing database load and improving API response times. - Maintained end-to-end type safety across the full stack using TypeScript, eliminating runtime errors in client-server data handling.

View project

Acme Store

Next.js, Stripe, Tailwind CSS, Zustand | 2026 - Integrated Stripe Checkout Sessions for secure, professional-grade payment flows with automated order confirmation via Stripe Webhooks syncing the database with payment events. - Built an SEO-friendly product storefront using Next.js Server Components with near-instant load times and persistent shopping cart via local storage using Zustand.

View project