h
hassan4702

Hassan

@hassan4702
5,0(1)

AI SaaS Developer building production apps with Nextjs, Python and LLMs

Pakistan
Inglese, Urdu, Hindi
Alcune informazioni sono riportate in lingua inglese.
Chi sono
I build complete AI-powered SaaS products end-to-end auth, payments, databases, background jobs, and LLM integrations that hold up under real load. Not just demos. Recent work: a multi-tenant job auto-apply platform built with Next.js, Python, Temporal, Postgres, Playwright, and OpenAI/Claude APIs. Handles thousands of users. Stack: Next.js, React, TypeScript, Python, FastAPI, Postgres, OpenAI, Claude, LangChain, RAG, Stripe. Send your idea , I'll tell you honestly if I can build it and what it costs.... Continua a leggere

Competenze

h
hassan4702
Hassan
offline • 

Consulta i miei servizi

Siti Web e software IA
I will build a scalable ai saas mvp with nextjs, openai and stripe
Sviluppo di chatbot IA
I will build a custom ai chatbot and rag assistant for your website

Esperienza lavorativa

Principle Engineer

Synergy Pakistan • Full time

Sep 2025 - Present8 mos

• Architecting a multi-tenant AI-powered job auto-apply SaaS platform serving thousands of users, spanning a Next.js web client, TypeScript FeathersJS REST/SSE API, and Python Temporal worker services on PostgreSQL. • Designed and implemented scalable Temporal workflows in Python for per-user application pipelines, using deterministic workflow IDs for idempotency, signal-driven reactivity, and child workflows for isolated per-job retries. • Built Playwright-based browser automation at scale with rotating residential proxies and per-user concurrency control, automating ATS submissions across Greenhouse, Lever, and other job-application platforms. • Engineered an AI-powered question-answering pipeline using OpenAI and Anthropic Claude APIs to dynamically respond to job-application forms, with per-job resume tailoring and PDF generation. • Implemented JobBERT-v2 (1024-dim) embeddings for semantic job matching, integrated into the job-bank sync pipeline with PostgreSQL and pgvector for efficient retrieval. • Deployed services on Kubernetes via Helm with KEDA-based autoscaling (2→10 worker replicas), Server-Sent Events backed by Postgres LISTEN for real-time UI updates, and Stripe integration with audit logging and RBAC on the API.

AI_Vision Clinic

AI / Full-Stack Engineer

AI Vision Clinic • Full time

Jul 2024 - Sep 20251 yr 2 mos

• Implemented back-end services using Python frameworks including FastAPI and Flask, ensuring scalable and efficient processing pipelines. • Applied RAG frameworks such as LlamaIndex, Ollama, and LM Studio to improve context-aware information retrieval and response generation using Elasticsearch as the vector database. • Developed and integrated a real-time uploading metrics system using WebSockets for live data processing. • Utilized task queues and caching with Huey, Redis, and Celery to manage asynchronous transcription workloads and improve performance. • Deployed asynchronous services using Uvicorn behind NGINX for secure, scalable, and production-ready ASGI architecture.

Care_Innovations

AI/Web developer

Care Innovations • Part time

Jun 2023 - Sep 20233 mos

• Spearheaded initial development of “Block Fund” (later my FYP), a blockchain crowdfunding platform, focusing on its Smart Contracts module. • Developed and tested a specialized cryptography module within an application for a Central Bank Digital Currency (CBDC) proof-of-concept, focusing on secure data handling. • Collaborated with senior developers on implementing and refining cryptographic protocols for the CBDC project, ensuring adherence to security best practices.

1 Recensioni
5,0

(1)
(0)
(0)
(0)
(0)
Valutazione dettagliata
  • Livello di comunicazione del venditore
    5
  • Consiglia a un amico
    5
  • Servizio conforme alla descrizione
    5
1-1 di 1 recensioni
Ordina per
Più rilevante
    R

    ronaldortega330

    US

    Stati Uniti

    5

    Amazing delivery!!!the freelancer did great job and deliver me at very short notice thanks

    2 giorni

    Tempo

    Utile?
    No