Skip to content
View clossal's full-sized avatar

Block or report clossal

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
clossal/README.md

Mehreen Munir

AI Engineer | AI Agents, Automation & System Architecture

I build practical AI and data products for teams that need dependable research, automation, and decision support. My focus is evidence-aware workflows, clear human review points, and systems that are useful in day-to-day operations.

What I build

  • Research and RAG workflows — document ingestion, retrieval, grounded answers, source traceability, and explicit insufficient-evidence handling.
  • Automation and orchestration — review-first process automation with clear hand-offs, audit trails, and safe boundaries around external actions.
  • Machine learning and forecasting prototypes — feature engineering, model evaluation, dashboards, and backtests using reproducible data.

Featured work

A local research workspace for permitted PDFs: page-level evidence, cross-source analysis, citation validation, insufficient-evidence handling, and reviewable research memory. The public repository is source-only and does not redistribute publisher documents, model weights, indexes, or private research records.

A runnable local workflow that classifies sample support records, retrieves FAQ sources, prepares grounded drafts, requires human review, and records decisions in SQLite. It intentionally does not send email or touch customer data.

A Streamlit demonstration using synthetic Solana-style data for feature engineering, a lightweight classification model, trend clustering, and a simple backtest. It is a portfolio prototype—not financial advice or a live signal.

An interactive Streamlit simulator for exploring strategy thresholds, risk controls, model signals, and backtest results on synthetic market-style data.

Demonstrated tools

Python, SQLite, Streamlit, scikit-learn, feature engineering, retrieval design, workflow state, review queues, audit logging, and local-first prototypes.

Ongoing work

The public research engine is available above. Prototype work on Reddit GTM research remains clearly marked as in progress; it is not presented as a finished agent system.

Working together

I am available for focused AI engineering work: internal document Q&A, research and analysis workflows, review-first automation, and data or forecasting prototypes.

  • Email Mehreen
  • Review each project README before assuming any feature is production-ready.

Popular repositories Loading

  1. time-series-projects time-series-projects Public

    Some noteworthy project I have done for different clients over the years. These projects data will not be shared due to privacy reasons.

    Jupyter Notebook

  2. Deep-Learning Deep-Learning Public

    Different projects related to deep learning, mostly RNN, LSTM, GRU, and others.

  3. solana-ml-dashboard solana-ml-dashboard Public

    Python

  4. Telegram-Strategy-Simulator Telegram-Strategy-Simulator Public

    Python

  5. reddit-gtm-research reddit-gtm-research Public

  6. customer-support-automation-human-review customer-support-automation-human-review Public

    Customer Support Automation & Human Review System — a review-first, traceable support workflow portfolio implementation.

    Python