Skip to content
View shedipes95's full-sized avatar

Block or report shedipes95

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
shedipes95/README.md

Sepideh Saadaei Jahromi

I work with data and AI: cleaning and analysing data, building and deploying models, and checking that the results hold up before anyone relies on them.

I finished an MSc in Computing (Human-Centred AI) at TU Dublin in 2026 with First Class Honours, after an earlier MSc in Financial and Risk Management Engineering. I'm based in Dublin and open to data, analytics and AI roles.

Projects

credit-threshold-lab · live app
Move the decision threshold on a credit-risk model and watch what changes: the confusion matrix, the cost of the mistakes, and who they fall on. Working the cost arithmetic backwards shows that the usual 0.5 cut-off assumes a missed default costs about 11 times more than a wrongly refused applicant. Tested functions, 37 tests, CI.

credit-default-nn-shap
A Keras network predicting loan default on 307,511 Home Credit applications, about 8% of them positive. Evaluation built around recall rather than accuracy, SHAP explanations for individual predictions, and an honest write-up of what the model gets wrong.

sql-finance-analytics
Analytical SQL on a synthetic retail-bank database: joins, CTEs and window functions answering business questions such as monthly flows, cohorts and default rates by segment.

MYAI-MVP · live demo
A React and TypeScript personal-finance dashboard from a team project at the National AI Challenge Ireland, where I built the front end: upload a bank CSV, see spending by category, and track budgets, savings and loans. The demo is front-end only: sign in with any email and password.

groundcheck
The checking code from my MSc thesis, cleaned up and published. It follows the FActScore approach (Min et al., 2023) to check an LLM's answer claim by claim against its source, and adds a measure of how well the automated verdicts agree with a human reviewer.

Also here: credit-scoring-fairness (credit model with SHAP and a fairness check across sex and age), market-risk-returns, financial-sentiment-nlp, and a few small web projects: tastely-frontend (live) with its Express proxy, and django-crud-htmx.

Tools

Python, SQL, JavaScript and TypeScript. pandas, scikit-learn, TensorFlow/Keras and SHAP for data and models. React, Next.js, FastAPI, Flask, Django and Express for apps and APIs. Git, pytest, GitHub Actions, Docker, Azure and AWS for shipping it. Power BI and Excel for reporting.

LinkedIn · shedipes95@gmail.com

Pinned Loading

  1. credit-threshold-lab credit-threshold-lab Public

    Move the decision threshold on a credit-risk model and watch what it costs — and who the mistakes land on

    Python

  2. credit-default-nn-shap credit-default-nn-shap Public

    Neural network for credit-default prediction on the Home Credit dataset — hyperparameter sweep, evaluation on imbalanced data, and SHAP explainability (global + per-case)

    Python

  3. sql-finance-analytics sql-finance-analytics Public

    Advanced SQL analytics on a synthetic Irish retail-bank dataset — JOINs, CTEs & window functions answering business questions (SQLite + Python)

    Python

  4. MYAI-MVP MYAI-MVP Public

    React + TypeScript personal-finance dashboard: upload a bank CSV, see spending by category, and track budgets, savings and loans

    TypeScript

  5. groundcheck groundcheck Public

    Audit whether an LLM's output is actually grounded in its source — atomic-claim verification, NLI cross-check, and agreement scoring against human labels

    Python

  6. credit-scoring-fairness credit-scoring-fairness Public

    End-to-end credit-risk model on the German Credit dataset — SHAP explainability + a fairness audit across sex and age (scikit-learn, SHAP)

    Python