Skip to content
View rambodbehmardi's full-sized avatar

Block or report rambodbehmardi

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. Markdown is supported. 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
rambodbehmardi/README.md

Hi, I'm Rambod.

I turn operational data into models, workflows, and decisions that can be traced and tested.

I work on data engineering and analytics at Miare, where I build across ClickHouse models, Python and SQL pipelines, Apache Superset semantic layers, and n8n decision workflows. I enjoy the point where a vague operational question becomes a reproducible data product—and where the final number has a visible grain, denominator, lineage, and limitation.

My background in industrial engineering, business analytics, and operations research means I tend to work across the whole decision path: from event history and data quality, through metric definition and experiment analysis, to the workflow or dashboard that someone actually uses.

What I work on at Miare

My work sits behind urban-logistics operations. The production details stay private, but the engineering patterns are consistent and reviewable:

Area What I build and verify
Event and metric models Version-aware histories, deduplication, final-event selection, late revisions, explicit grains, and maturity-aware denominators.
Decision workflows Stateful n8n flows with configuration validation, independent guardrails, keyed merges, diff-only updates, read-back verification, reporting, and hard failure paths.
Policy and experiment analysis Event-time treatment reconstruction, execution-fidelity audits, difference-in-differences, bootstrap uncertainty, heterogeneity checks, and a clear line between descriptive and causal evidence.
Analytics release engineering ClickHouse query QA and reproducible Superset dataset/dashboard bundles with structural validation before handoff.
Reconciliation and controls Ledger components, adjustments and reversals kept separate from operational rates, with explicit invariants and explainable exceptions.

The public workflow catalog maps 18 operational analytics themes to clean-room patterns without exposing production SQL, schemas, identifiers, thresholds, endpoints, or decision rules.

Featured work

GDELT event intelligence · graduate research project

For my graduate research project at Sharif University of Technology, I designed and implemented a computational-social-science platform for near-real-time analysis of high-volume daily GDELT event streams. The original system used SQL Server 2019, SSIS, SSAS Tabular, DAX, and bilingual Power BI dashboards for global-event, event-type, country, tone/news, and actor-network views.

I rebuilt the core engineering ideas as a public, portable lab:

generated events → immutable observations → validation/rejection → version-resolved facts → dimensional model → named SQL → reconciliation

The GDELT Event Intelligence Platform includes ten tests, five documented analytical queries, deterministic replay, revision handling, rejected-record auditing, and a transparent mapping from the original research stack to the public reconstruction.

Urban logistics data platform · production patterns, reconstructed safely

Operational data becomes difficult when an event can arrive twice, change later, or mean something different under the policy that was active at event time. I built the Urban Logistics Data Platform to make those problems inspectable with generated data.

The runnable vertical slice covers immutable evidence, canonicalization, late-arriving revisions, analytical marts, ledger reconciliation, geospatial plausibility, event-time switchback reconstruction, 29 executable contracts, and 10 tests. It is the public engineering proof behind the 18-pattern workflow catalog—not a copy of a production system.

Circular supply-chain optimisation · working paper

My 2026 working-paper overview develops a scenario-based stochastic multi-objective MILP for a multi-product, multi-period closed-loop supply chain. Disruption uncertainty is represented through probability-weighted scenarios; return-rate variation is studied through sensitivity analysis. The research implementation is written in Python, and the analysis examines how Pareto-optimal trade-offs move across configurations.

The repository is deliberately honest about its current boundary: it publishes the research design, not model code or results that are not yet part of a reproducibility package.

More engineering labs

Project What it demonstrates Reproducible proof
Clinical Analytics Privacy Lab FHIR-like generated resources, immutable raw storage, typed staging, quality gates, k-suppressed marts, and a privacy audit; with optional dbt/Dagster reference paths. 7 tests and CI
Equine Ecosystem Analytics Multi-tenant operational modelling, composite tenant keys, append-only product events, analytical marts, reconciliation, and tenant-scoped retrieval. 7 tests and CI
Urban Logistics Route Optimizer Deterministic capacitated routing with sweep clustering, nearest-neighbour construction, 2-opt improvement, feasibility checks, and distance/load reconciliation. 8 tests and CI

These are independent engineering labs built with generated data. They demonstrate how I approach unfamiliar domains; they are not claims of production experience in clinical or equine systems.

Technical range, in context

  • Operational analytics: ClickHouse, SQL, Python, JavaScript in n8n, n8n orchestration, Apache Superset, metric contracts, and release validation.
  • Research data platforms: SQL Server 2019, SSIS, SSAS Tabular, DAX, Power BI, dimensional modelling, and high-volume event data.
  • Public lab stack: Python, SQLite, dbt and Dagster reference paths, Docker Compose, deterministic test data, and GitHub Actions.
  • Decision methods: event-time analysis, controlled experiments, reconciliation, sensitivity analysis, DEA, and multi-objective MILP.

How I build

  • I define the grain and denominator before I defend a metric.
  • I keep raw evidence separate from corrected and analytical states.
  • I design the failure path, read-back check, and rollback boundary with the happy path.
  • I use generated examples to make production patterns reviewable without exposing production systems.
  • I would rather label a result inconclusive than turn uncertainty into a confident claim.

Education and research

  • M.Sc. in Business Administration, Strategic Management & Business Analytics — Sharif University of Technology, ranked 2nd in cohort, full merit scholarship.
  • B.Sc. in Industrial Engineering — Sharif University of Technology, ranked 10th of 470, full merit scholarship.
  • Research profile: ORCID 0009-0003-8035-2611.

Every public repository here states its data origin, proof, and limitations. Production data and proprietary logic stay private; runnable implementations, tests, contracts, and engineering reasoning do not.

Popular repositories Loading

  1. milp-reverse-logistics-circular-economy milp-reverse-logistics-circular-economy Public

    Working-paper overview of scenario-based multi-objective MILP for a sustainable closed-loop supply chain

    1

  2. rambodbehmardi rambodbehmardi Public

    Data engineering, analytics and decision systems—from urban logistics to open-data research

  3. urban-logistics-data-platform urban-logistics-data-platform Public

    Synthetic event pipeline with idempotent ingestion, analytical marts, data contracts and observability

    Python

  4. clinical-analytics-privacy-lab clinical-analytics-privacy-lab Public

    Synthetic clinical-style pipeline with quality gates, privacy suppression, dbt and Dagster reference paths

    Python

  5. equine-ecosystem-analytics equine-ecosystem-analytics Public

    Synthetic multi-tenant product analytics with event history, tested marts and scoped retrieval

    Python

  6. urban-logistics-route-optimizer urban-logistics-route-optimizer Public

    Deterministic capacitated routing baseline with generated instances and feasibility tests

    Python