I identify operational problems, understand the existing workflow, redesign inefficient processes, and build practical systems using automation, APIs, internal tools and AI-assisted implementation.
My focus is on reducing unnecessary manual work, making workflows easier to follow and designing for more consistent output. AI helps accelerate implementation; problem definition, system decisions, review and validation guide the work.
Real operational case study · Delivered and used operationally
An internal business system for Behrad Gas connecting operational information, invoice workflows and reporting. Delivered and used operationally, with ongoing supervision and iterative improvements.
I led problem discovery, requirements, workflow design, product decisions, iterative validation and delivery. Implementation was AI-assisted and developed under my direction, review and testing.
Explore the operational case study — public documentation, conceptual diagrams and clearly labeled synthetic examples; private implementation and company data are excluded.
Working, field-tested MVP · Documentation-only case study
Designed and iterated a browser-based multilingual conversation workflow for face-to-face use, including explicit turn-taking, field testing under noisy conditions, and documented product trade-offs.
Explore the PyxMint case study — conversation workflow, conceptual diagrams, qualitative café observations and a proposed test matrix. Application source, real transcripts and private operational data are excluded. Field observations are qualitative, not benchmarks or evidence of equal performance across languages.
Runnable Python implementation · Tests and CI
A repeatable reporting workflow that validates CSV/Excel input, separates accepted and rejected records, and produces a six-sheet workbook. Documentation explains rejection rules, metric meanings and source-row limitations.
Explore the reporting implementation — runnable example, automated tests, CI and synthetic report evidence.
Focused Python implementation · Tests and CI
A supporting data-preparation workflow that normalizes headings and text, removes exact duplicates after normalization, checks required-column presence and reports remaining missing values.
Explore the cleaner implementation — runnable example, automated tests, CI and synthetic before/after evidence. Semantic email/date validation and an integration API are outside the current implementation.
Understand the problem → map the existing workflow → identify friction and manual work → design the system → implement/integrate → test and iterate → document and deliver.
| Capability | Public evidence | Scope / limits |
|---|---|---|
| Requirements, workflow design and operational delivery | Industrial case study | Delivered internal system; documentation and synthetic examples, not public application code or measured business outcomes. |
| Product decisions and testing under real conditions | PyxMint case study | Documentation-only account of a field-tested MVP; qualitative observations, with future tests clearly labeled. |
| Data validation and reporting automation | Reporting implementation | Runnable Python/Pandas/openpyxl workflow with tests and CI; local file processing, not a deployed reporting service. |
| Data normalization and preprocessing | Cleaner implementation | Supporting Python/Pandas implementation with tests and CI; required-column checks, not comprehensive semantic validation. |
These repositories demonstrate different kinds of evidence. The implementation examples do not establish public FastAPI or API-service implementation evidence. Tests and CI cover the included scenarios; they are not comprehensive reliability guarantees.
For portfolio examples and a discussion of an operational workflow or practical system: