Ikram K.

Top Rated AI Engineer with 100% Job Success. I build production LLM and machine learning applications: RAG systems, AI agents, custom chatbots, and fine-tuned models, deployed on AWS with Python and Rust. 18 jobs completed, every rating 5⭐. Clients hire me when a generative AI demo works in a notebook but needs to become a reliable product: accurate on real documents, fast at scale, secure with client data, and cheap to run. WHAT I CAN BUILD FOR YOU • AI agents and chatbots: Tool-using agents and customer-facing chatbots built with LangChain, DSPy, OpenAI, Anthropic Claude, or local open-source models, with memory, guardrails, and logging so you can see what the model did and why. • RAG and document AI: Chat over PDFs, contracts, product catalogs, and internal knowledge bases using hybrid search (keyword + vector), tuned chunking, reranking, and an evaluation suite so accuracy is measured, not guessed. Built a retail apparel compliance checker that reached 98% accuracy and replaced manual verification. • LLM fine-tuning: Fine-tuned Vision-Language models on a 144K-image dataset on Lambda Labs GPUs and deployed it on Hugging Face. Prepared the dataset, used Hugging Face AutoTrain, applied LoRA/PEFT, and optimized evaluation and inference for open-source models (LLaMA, Gemini, Qwen). • Secure private deployments: Multi-tenant B2B knowledge hubs with strict per-client document isolation and local LLMs, on Django backends. Recently delivered one successfully for a B2B client under a $15K+ contract. • ML and MLOps on AWS: SageMaker training and inference pipelines, XGBoost and PyTorch models, Docker, serverless, CI/CD, and monitoring. Certified in Data Engineering (DeepLearning.AI), RAG (DeepLearning.AI), and LLMOps (Duke University). WHAT CLIENTS SAY "Ikram was fully committed from day one and always gave more than 100%..." - Secure B2B knowledge Hub, client "Working with Ikram was 10/10 great once again. Looking forward to continuing to work together!" - RAG compliance client, repeat contract TECH STACK Python, FastAPI, LangChain, Django, TypeScript, React, Rust, Tauri (Desktop app), AWS (SageMaker, Lambda, S3), Docker, Pinecone, Qdrant, Milvus, PostgreSQL, PyTorch, Scikit-learn, XGBoost Depth beyond the frameworks: I implemented a GPT-style LLM training and inference pipeline from scratch in Rust (a port of Karpathy's nanochat), so I can debug what the libraries hide. HOW WE START Send a message with what you are building and, if possible, a few sample documents or a data sample. I reply within hours with a concrete plan, a realistic accuracy target, and a fixed price or hourly estimate. Available 30+ hrs/week and open to contract-to-hire.