Available for projects Remote worldwide
Ibadat Ali AI Engineer & AI Agent Developer
I engineer AI agents and AI-powered products — connecting models, tools, retrieval, APIs, and evaluation into systems people can understand and operate.
Technologies and tools shown across the selected project catalog: WhatsApp / Baileys, Node.js, Python, FastAPI, Tesseract OCR, Groq Vision, n8n, Stripe API, Docker, DigitalOcean, Next.js 15, TypeScript, Tailwind CSS, Prisma, PostgreSQL, role-based dashboards, Prophet, XGBoost, SARIMA, LSTM, React, Vite, Chart.js, Hugging Face, PyTorch, TCN, Isolation Forest, Autoencoder, SHAP, PuLP, React 18, Tailwind, Radix UI, Zustand, SQLAlchemy, Gemini 2.0 Flash, Google Gemini, Google Maps, Gmail, Google Sheets, AI agent, web scraping, MCP, Tree-sitter, sentence-transformers, Chroma, CLI, WhatsApp Business, Simple memory, FAQ + orders + inventory, MySQL, depthwise-separable CNN, Coordinate Attention, persistent homology, GUDHI, EfficientNet-B0, Grad-CAM, Flask, jsPDF, pandas, scikit-learn, Random Forest, Gradient Boosting, Jupyter, SQL, SQLite, data visualization, Matplotlib, seaborn.
How I build One connected process
Understand the work, build around its constraints, then hand over something people can use and maintain.
Explore the delivery approachMeet Ibadat 01:14
A personal introduction to my AI engineering work, agent development, and approach to building dependable systems.
A muted preview may start when you scroll here. Use the player controls to pause, replay, or enable sound.
01 / Selected projects
Explore agent workflows, developer tools, model-backed products, and applied ML systems. Each case study distinguishes its scope, status, implementation approach, and available evidence.
WhatsApp intake · OCR · duplicate detection · Stripe verification
Result — Delivered payment-verification workflow with queueing, OCR fallback, Stripe reconciliation, duplicate handling, and handoff documentation.
Education · Client platform
Result — Production client platform with documented handoff and a live website.
Demand forecasting · 4-model ensemble
Result — Four-model comparison with live inference and downloadable planning reports.
Predictive maintenance · anomaly detection + scheduling
Result — NASA CMAPSS RUL study with anomaly detection, SHAP attribution, and scheduling.
Multimodal clinical workflow · Gemma 4 Hackathon
Result — Gemma hackathon submission with multimodal inputs, audit logging, and offline fallback.
Capture · enrich · outreach · report
Result — Client workflow proof spanning capture, enrichment, outreach, and reporting.
Agent tool-calling · MCP & RAG
Result — Open-source MCP tooling for structured repository context before agent actions.
WhatsApp trigger · memory · tool use
Result — Client workflow proof connecting WhatsApp intake, memory, FAQ, order capture, and inventory tools.
Automated ETL · global analytics
Result — Database-backed analytics platform connecting automated ETL, APIs, and an interactive dashboard.
Topology-aware KD research
Result — Research prototype exploring topology-aware knowledge distillation for compact CT classification.
Multimodal assisted-review prototype
Result — Assisted-review prototype combining image, audio, and vital-sign inputs in one bounded workflow.
Deployable price-prediction workflow
Result — Deployable inference workflow translating e-commerce transaction data into price estimates.
Procurement analytics workflow
Result — Procurement analysis workflow connecting sales, purchases, pricing, invoices, and vendor performance.
Exploratory data analysis
02 / Capabilities and delivery
Agent architecture and AI engineering, from model and tool integration through APIs, evaluation, and a documented handoff.
01 / Agent development
Build tool-calling agents, MCP integrations, retrieval workflows, and AI automations with clear boundaries and dependable validation.
Tools and methods
02 / AI engineering
Engineer deep-learning, anomaly-detection, and retrieval systems around repeatable experiments, explainability, and practical use cases.
Tools and methods
03 / Modeling and evaluation
Build and compare predictive approaches with interpretable features, forecasting methods, and clear evaluation boundaries.
Tools and methods
04 / Data foundations
Prepare, query, and inspect structured data so AI and machine-learning workflows start from clear, usable evidence.
Tools and methods
How I work
Project scopes are agreed before implementation; delivery is planned around the work, constraints, and handoff needs.
03 / Evidence
I make project scope and evidence easy to distinguish, so a demo, research build, or client delivery is not presented as something it isn’t.
Each catalog entry opens a case study with its problem, approach, result, technology stack, and links where available.
Client work, live builds, research, prototypes, and labs are labeled separately for context.
Private client work is shown without source links or confidential implementation details.
04 / About
From dataI’m Ibadat, an AI Engineer and AI Agent Developer based in Pakistan. I build tool-using agents, retrieval workflows, model-backed APIs, and applied ML products.
I focus on clear system boundaries: what the model decides, what tools can do, how outputs are checked, and how people can operate the result.
Read the résumé05 / Experience
Industry roles and independent delivery across ML engineering, LLM-integrated systems, and AI agent workflows.
2025 — Present (paused Dec 2025 – Mar 2026 for full-time role)
Fiverr · Remote
Delivering predictive modeling, forecasting, NLP, RAG, computer vision, and AI automation systems for international clients, with stakeholder-ready reports and structured handoff documentation.
Dec 2025 — Mar 2026
Soft Shack · Full-time, on-site
Designed production ML and LLM-integrated automation pipelines with FastAPI and Docker, adding RAG retrieval, confidence gates, audit logging, and client-ready technical handoffs.
May 2025 — Aug 2025
Arch Technologies · Internship
Built and evaluated forecasting and predictive analytics models with Python, Scikit-learn, XGBoost, and Pandas; engineered SQL/CSV ETL workflows and delivered SHAP-based reports.
Education and development
Self-directed learning in machine learning, AI systems, and software engineering — strengthened through client delivery, applied research, and focused technical certifications.
Credentials
Open certificate
Open certificate
06 / Contact
Share the context, constraints, and what a useful outcome would look like. I’ll reply within one business day.
Most projects begin with a focused 30-minute conversation about the problem, constraints, and fit.
Signature / 01
— Ibadat Ali