10/11/2026 updated
partially available
info: Available evenings and weekends
Python & FastAPI Developer | AI Agents & Automation
sargodha, Pakistan
Only remote
BS Artificial Intelligence (expected 2028)About me
AI developer building Python/FastAPI backends, LLM-powered applications, RAG systems, and workflow automations. Experienced with REST APIs, Docker, React, and third-party integrations through hands-on projects and an ML internship. Open to remote, project-based work.
WorkflowsWorkflow Management SystemsMulti-Agent SystemsAutomation Anywhere
AI Agent & LLM Application Development
Building AI agents, RAG applications, and lead qualification tools using Python, LangGraph, LangChain, and structured LLM outputs combined with deterministic rules and edge-case testing.
Workflow Automation
Development of end-to-end business workflow automation pipelines, including caption generation, HTML/CSS post rendering with Playwright, and integration with external APIs such as the Meta Graph API.
Machine Learning & Model Evaluation
Analyzing large-scale data, comparing classification models such as Logistic Regression, Decision Tree, and Random Forest using client-holdout validation, and evaluating ranking models against baselines with a focus on leakage control and business-aligned metrics.
FastAPI & REST API Development
Building and consuming REST APIs using FastAPI and TypeScript, with experience in structured outputs and provider rate-limit handling.
SQL & Data Analysis
Working with SQL for data querying and analysis, including investigation of search and engagement data across large page sets.
Pydantic & Data Validation
Applying Pydantic for data validation within LLM pipelines, ensuring reliable and structured outputs from AI models.
Concurrency & Performance Testing
Running concurrency tests to identify latency issues and provider rate-limit failures in AI-driven qualification pipelines.
Docker & Git
Using Docker for containerization and Git for version control in software development workflows.
MCP & LangGraph
Utilizing MCP and LangGraph frameworks for building multi-step AI agent workflows and orchestration pipelines.
Building AI agents, RAG applications, and lead qualification tools using Python, LangGraph, LangChain, and structured LLM outputs combined with deterministic rules and edge-case testing.
Workflow Automation
Development of end-to-end business workflow automation pipelines, including caption generation, HTML/CSS post rendering with Playwright, and integration with external APIs such as the Meta Graph API.
Machine Learning & Model Evaluation
Analyzing large-scale data, comparing classification models such as Logistic Regression, Decision Tree, and Random Forest using client-holdout validation, and evaluating ranking models against baselines with a focus on leakage control and business-aligned metrics.
FastAPI & REST API Development
Building and consuming REST APIs using FastAPI and TypeScript, with experience in structured outputs and provider rate-limit handling.
SQL & Data Analysis
Working with SQL for data querying and analysis, including investigation of search and engagement data across large page sets.
Pydantic & Data Validation
Applying Pydantic for data validation within LLM pipelines, ensuring reliable and structured outputs from AI models.
Concurrency & Performance Testing
Running concurrency tests to identify latency issues and provider rate-limit failures in AI-driven qualification pipelines.
Docker & Git
Using Docker for containerization and Git for version control in software development workflows.
MCP & LangGraph
Utilizing MCP and LangGraph frameworks for building multi-step AI agent workflows and orchestration pipelines.
Languages
EnglishFluent
Project history
Built an inbound lead qualification pipeline combining seven LLM-extracted signals, Pydantic validation, and deterministic scoring rules. Tested prompt injection, missing fields, ambiguous intent, and buyer-versus-distributor cases; ran concurrency tests to identify latency and provider rate-limit failures.
Built a workflow for caption generation, HTML/CSS post rendering with Playwright, and approval or rejection through WhatsApp before publishing via the Meta Graph API.
Analyzed search and engagement data across 30,000 pages to investigate declining performance and prioritize improvement opportunities. Compared Logistic Regression, Decision Tree, and Random Forest models using client-holdout validation; Random Forest achieved ROC AUC 0.747 and precision at 50 of 0.68. Evaluated ranking models against a simple baseline using captured missed clicks, highlighting the importance of leakage control and business-aligned metrics.