You will get an AI agent reliability audit with an actionable test plan
Project details
I’ll review an existing AI agent, RAG workflow, or LLM-backed product and map the failures most likely to reach users. You’ll receive a review of prompts, tool calls, retrieval, and fallback behavior, a focused eval plan with high-value test cases, a written findings report ranked by impact and effort, and a 30-minute walkthrough. This works best for teams that already have a prototype or production workflow and want clear next steps before investing in a larger rebuild.
AI Algorithms
Large Language Model, Transformer ModelAI Applications
AI Chatbot, Conversational AI, Natural Language UnderstandingAI Development Language
PythonAI Models
ChatGPT, GPT-4What's included $350
These options are included with the project scope.
$350
- Delivery Time 5 days
- Number of Revisions 1
- Model Documentation
- Model Testing & Optimization
- Natural Language Processing
- Prompt Engineering
Frequently asked questions
About Aaliyan
Production AI Agent Engineer | RAG, Evals, Python, TypeScript
Karachi, Pakistan - 10:30 am local time
Recent work includes a bilingual Gemini support agent that calls live vehicle, payment, maintenance, customer, and ticket tools, escalates cases safely, handles WhatsApp media, and runs against repeatable eval scenarios.
Selected production work
- Cut document-processing cost by 20x while improving extraction quality.
- Reduced an AI workflow's runtime by 3x through parallel execution and retry design.
- Built evals for ticket intent, required-field collection, confirmation, payload validity, media handling, and unsafe promises.
- Shipped real-time voice and chat flows with Twilio, WebSockets, PostgreSQL, Redis, Docker, and AWS.
Typical project scope
- Custom AI agents in Python or TypeScript.
- RAG with grounded answers, citations, confidence checks, and abstention.
- Tool calling with failure handling, retries, timeouts, and payload validation.
- LLM evals and regression tests.
- OCR and document automation.
- Backend services and admin interfaces around AI workflows.
My working stack includes Python, FastAPI, Django, TypeScript, React, Mastra, LangGraph, Gemini, OpenAI, Claude, PyTorch, Hugging Face, PostgreSQL, Redis, Docker, AWS, FAISS, and Weaviate.
I usually begin with a small, testable milestone that makes the system's failure modes visible before expanding the build.
Steps for completing your project
After purchasing the project, send requirements so Aaliyan can start the project.
Delivery time starts when Aaliyan receives requirements from you.
Aaliyan works on your project following the steps below.
Revisions may occur after the delivery date.
Review the current system
I’ll inspect the agent flow, prompts, retrieval, tools, fallbacks, and available logs or evals.
Build the eval and failure map
I’ll identify high-risk failure modes and define focused test cases around the most important user flows.