Shafi U

Full Stack AI Engineer

Cognilium AI • Temps plein

Feb 2025 - Present • 1 yr 3 mos

At Cognilium AI, I operate as a Full Stack AI Engineer, specializing in architecting production-ready AI/ML systems, complex Retrieval-Augmented Generation (RAG) pipelines incorporating advanced techniques like HyDE, and highly scalable cloud solutions. My core focus is engineering secure, enterprise-grade applications from the ground up, leveraging a modern stack that includes Python, TypeScript, Next.js, React, LangChain, and advanced vector databases like Milvus and AstraDB. I deploy these systems across AWS ECS, GCP Cloud Run, and Azure Container Apps using Docker containerization and Terraform for infrastructure as code. A defining project during my tenure is FamilyOffice AI, a robust FinTech SaaS platform. I architected its zero-trust, multi-tenant infrastructure using Next.js, React, FastAPI, and Firestore on GCP Cloud Run. To drive its core intelligence, I implemented an advanced AI Supervisor Agent powered by Google ADK and Gemini 2.0 Flash, enabling autonomous and secure financial data processing for high-end users. In the legal tech sector, I spearheaded the technical design for LegalLens, an enterprise Contract AI SaaS. I engineered the complete full-stack architecture using FastAPI, React, and PostgreSQL, ensuring strict security and data integrity. A major highlight was building its highly scalable semantic search engine; I integrated Azure OpenAI GPT-4.1 with a Milvus vector database to execute instantaneous, deep-context contract analysis. Beyond these flagship platforms, I build specialized, autonomous agents such as VectorHire for AI-driven recruitment—leveraging Ultravox and Twilio for autonomous voice interviews—and ProspectVox for B2B lead generation. I also design complex enterprise AI automation workflows utilizing n8n, LangChain, OpenRouter, LiteLLM Server, and AWS Bedrock Agentcore, transforming manual processes into self-monitoring, intelligent systems.

Python Backend Developer

IDEEZA • Temps plein

Feb 2024 - Apr 2025 • 1 yr 2 mos

At IDEEZA, I served as a Python Backend Developer, focusing on engineering high-performance, scalable backend architectures and establishing robust CI/CD deployment pipelines. My primary objective was to build resilient server-side infrastructure capable of handling complex, data-heavy workflows and seamless third-party integrations. I leveraged a core technology stack encompassing Python, FastAPI, Django, PostgreSQL, and Docker, ensuring all applications were heavily optimized for speed, security, and reliability within enterprise Linux environments. A major pillar of my work involved designing and implementing advanced RESTful APIs. By optimizing the underlying architectural logic and database queries within FastAPI and Django, I successfully reduced overall API response times by 30%. To guarantee continuous system stability and rapid iteration, I architected comprehensive CI/CD pipelines utilizing GitHub Actions. This established strict automated testing and deployment protocols that minimized production downtime and significantly accelerated the delivery of new features. To drive the company's data operations, I engineered highly efficient, automated web scraping and data extraction solutions utilizing BeautifulSoup, Scrapy, and Selenium. Furthermore, I designed sophisticated, event-driven workflows to handle intensive background processing and seamlessly manage complex third-party webhook integrations, including secure payment routing via Stripe and automated communications via Twilio. Additionally, I expanded IDEEZA's automation capabilities by developing scalable, real-time chatbots utilizing Python-Telegram-Bot and WebSockets. By seamlessly bridging these conversational interfaces with our core backend databases and APIs, I delivered end-to-end automation solutions that actively streamlined user interactions and drastically reduced manual operational overhead.

Python Developer Intern

Signature Technologies • Temps plein

Jun 2022 - Sep 2022 • 3 mos

Visa Application Automation System: Engineered a robust Python automation suite using Selenium and Pandas to streamline the Australian visa application process. Developed an intelligent bot capable of parsing bulk user data from Excel files and autonomously filling complex 16-page application forms with high accuracy. Implemented automated login handling, real-time logging, and status monitoring to ensure reliability. Integrated the Twilio API to trigger instant SMS notifications to agents upon successful submission, significantly reducing manual data entry time and errors.