Touseef Ahmad

Junior AI Engineer

• Job Details:Collaborated directly with company CEOs to develop an AI-powered mobile application, leading all backend and AI-related development including API creation and integration. Built and deployed a modular FastAPI backend, implementing structured conversation flows, handwriting recognition, and audio-based feedback using OpenAI GPT-4 and Whisper. Developed image analysis pipelines to validate handwritten Urdu alphabets, returning adaptive encouragement or retry messages through AI-generated speech. Engineered conversation tracking with dynamic response types (e.g., tracing or audio), including session-based UUID generation and context preservation. Containerized and deployed services using Docker Compose with PostgreSQL. Spearheaded a skin check AI project by modifying model architecture to support base64 image inputs, and successfully deployed it on Vertex AI for scalable real-time inference. Contributed to open-source multi-agentic frameworks, notably Hugging Face’s SmolAgents and Orchestra, with multiple recognized contributions that enhanced agent orchestration for complex tasks.

AI/ML Python Developer Intern

• Job Details:Fine-tuned small language models (SLMs) using Hugging Face GGUF models, employing prompt engineering and PEFT for optimized query handling and task performance. Built and deployed Skillsphere’s backend, integrating RAG pipelines and OpenAI assistants to deliver AI-driven insights and intelligent query responses. Designed and implemented a log analyzer tool using Python and Streamlit, capable of identifying suspicious system activity and generating actionable security insights. Developed a finance AI agent using phi-data, enabling financial trend analysis, prediction modeling, and reporting automation. Engineered an agentic RAG system for document (PDF) processing, incorporating real-time web browsing for enriched contextual understanding. Created an AI developer assistant agent for code debugging, optimization, and automatic retrieval of GitHub issues and documentation.